make style

revert make file
fix style
2021-09-13 10:51:52 -07:00 · 2021-09-13 18:42:10 +03:00 · 2021-09-13 18:41:43 +03:00 · 2021-09-13 11:39:22 +03:00 · 2021-09-13 11:34:44 +03:00 · 2021-09-03 11:59:51 +03:00
350 changed files with 27201 additions and 44152 deletions
@@ -1,12 +0,0 @@
-# extends https://github.com/jupyterhub/repo2docker/blob/main/repo2docker/buildpacks/conda/environment.yml
-# see https://mybinder.readthedocs.io/en/latest/using/config_files.html#environment-yml-install-a-conda-environment
-channels:
- conda-forge
-dependencies:
- xeus-cling=0.13.0
- xwidgets=0.26.0
-# NOTE: it's possible these aren't needed for the lab frontend
- widgetsnbextension=3.5.1
- pip
- pip:
-  - glvis==0.3.2
@@ -1,26 +0,0 @@
-#!/bin/bash
-
-set -e
-
-# cling is installed here (in bin) and will look in {dir}/include and {dir}/lib
-# without extra intervention (jk it doesn't look in {dir}/lib unless something
-# has been #included from {dir}/include first...)
-install_dir=/srv/conda/envs/notebook
-mkdir -p $install_dir
-
-# build and install mfem, which is the directory we start in
-make serial SHARED=YES -j8
-make install PREFIX=$install_dir
-
-# install xeus-glvis
-git clone https://github.com/GLVis/xeus-glvis.git
-pushd xeus-glvis
-make install prefix=$install_dir
-popd
-
-# install jupyter-lab extension
-jupyter labextension install @jupyter-widgets/jupyterlab-manager --no-build
-jupyter labextension install glvis-jupyter
-
-# fixup kernelspec, we could probably do this from sh but ¯\_(ツ)_/¯
-python .binder/update_kernel_env.py
@@ -1,14 +0,0 @@
-# Update the LD_LIBRARY_PATH of the C++14 kernel so it can find mfem without
-# extra pragma cling statements
-
-import json
-
-kernelspec = "/srv/conda/envs/notebook/share/jupyter/kernels/xcpp14/kernel.json"
-
-with open(kernelspec, "r") as f:
-    obj = json.load(f)
-
-obj["env"] = {"LD_LIBRARY_PATH": "/srv/conda/envs/notebook/lib"}
-
-with open(kernelspec, "w") as f:
-    json.dump(obj, f)
@@ -82,9 +82,9 @@ jobs:
      uses: mfem/github-actions/build-mfem@v2.0
      with:
        os: ${{ runner.os }}
-        target: opt
+        target: optim
        codecov: NO
-        mpi: par
+        mpi: parallel
        build-system: make
        hypre-dir: ${{ env.HYPRE_TOP_DIR }}
        metis-dir: ${{ env.METIS_TOP_DIR }}
@@ -63,7 +63,7 @@ jobs:
        exit 1

  code-style:
-    runs-on: ubuntu-18.04
+    runs-on: ubuntu-16.04 # needed for astyle 2.05.1

    steps:
    - name: checkout mfem
@@ -71,7 +71,7 @@ jobs:

    - name: get astyle
      run: |
-        sudo apt-get install astyle=3.1-1ubuntu2
+        sudo apt-get install astyle=2.05.1-0ubuntu1

    - name: style check
      run: |
@@ -105,9 +105,6 @@ jobs:

    - name: branch-history
      run: |
-        # We override origin to make sure we point to the main repo.
-        # This is to have consistent test results on PRs from forks.
-        git remote remove origin
-        git remote add origin https://github.com/mfem/mfem.git
+        git fetch origin master:master
        git checkout -b gh-actions-branch-history
        ./config/githooks/pre-push --history
@@ -51,8 +51,6 @@ examples/ex1[04-9]
 examples/ex1[0-9]p
 examples/ex2[0-9]
 examples/ex2[0-9]p
-examples/ex30
-examples/ex30p

 examples/refined.mesh
 examples/displaced.mesh
@@ -225,14 +223,6 @@ miniapps/mtop/ParHeat*
 miniapps/mtop/seqheat
 miniapps/mtop/SeqHeat*

-miniapps/autodiff/paradiff
-miniapps/autodiff/seqadiff
-miniapps/autodiff/seqtest
-miniapps/autodiff/par_example
-miniapps/autodiff/seq_example
-miniapps/autodiff/seq_test
-miniapps/autodiff/Exampl*
-
 miniapps/navier/navier_mms
 miniapps/navier/navier_kovasznay
 miniapps/navier/navier_kovasznay_vs
@@ -259,8 +249,6 @@ miniapps/performance/sol.*

 miniapps/shifted/distance
 miniapps/shifted/ParaViewDistance
-miniapps/shifted/extrapolate
-miniapps/shifted/ParaViewExtrapolate
 miniapps/shifted/diffusion
 miniapps/shifted/diffusion.mesh
 miniapps/shifted/diffusion.gf
@@ -300,15 +288,10 @@ miniapps/solvers/ParaView
 miniapps/solvers/mesh.*
 miniapps/solvers/sol.*

-miniapps/parelag/MultilevelHcurlHdivSolver
-miniapps/parelag/*.mesh
-
 # Unit test binary and outputs
 tests/unit/output_meshes
 tests/unit/unit_tests
 tests/unit/punit_tests
-tests/unit/cunit_tests
-tests/unit/pcunit_tests
 tests/unit/sedov_tests_*
 tests/unit/psedov_tests_*
 tests/unit/tmop_pa_tests_*
@@ -316,12 +299,6 @@ tests/unit/ptmop_pa_tests_*
 tests/unit/ceed_tests
 tests/unit/debug_device_tests

-# Benchmark binaries
-tests/benchmarks/bench_ceed
-tests/benchmarks/bench_tmop
-tests/benchmarks/bench_vector
-tests/benchmarks/bench_virtuals
-
 # Test script output
 tests/scripts/*.err
 tests/scripts/*.out
@@ -338,6 +315,3 @@ build-*/*
 # PETSc automated build
 petsc-build/*
 pkg.gitcommit
-
-# Jupyter Notebook Checkpoints
-.ipynb_checkpoints
@@ -13,84 +13,243 @@
 # at Lawrence Livermore National Laboratory (LLNL). This entire pipeline is
 # LLNL-specific!

+# We define the following GitLab pipeline variables:
+#
+# BUILD_ROOT:
+# The path to the shared resources between all jobs. For example, external
+# repositories like 'tests' and 'tpls' are cloned here. Also, 'tpls' is built
+# once for all targets, so that build happen here. The BUILD_ROOT is unique to
+# the pipeline, preventing any form of concurrency with other pipelines. This
+# also means that the BUILD_ROOT directory will never be cleaned.
+# TODO: add a clean-up mechanism
+#
+# REBASELINE:
+# Defines the default choice for updating the saved baseline results. By default
+# the baseline can only be updated from the master branch. This variable offers
+# the option to manually ask for rebaselining from another branch if necessary.
+#
+# MFEM_ALLOC_NAME:
+# On LLNL's quartz, there is only one allocation shared among jobs in order to
+# save time and resources. This allocation has to be uniquely named so that we
+# are sure to retrieve it.
+#
+# TPLS_REPO & TESTS_REPO:
+# Git repositories used in the pipeline
+#
+# ARTIFACTS_DIR:
+# Directory used to place artifacts.
+
+variables:
+  BUILD_ROOT: ${CI_BUILDS_DIR}/MFEM/${CI_PROJECT_NAME}_${CI_COMMIT_REF_SLUG}_${CI_PIPELINE_ID}
+  AUTOTEST_ROOT: ${CI_BUILDS_DIR}/MFEM
+  REBASELINE: "NO"
+  AUTOTEST: "NO"
+  ALLOC_NAME: ${CI_PROJECT_NAME}_ci_${CI_PIPELINE_ID}
+  TPLS_REPO: ssh://git@mybitbucket.llnl.gov:7999/mfem/tpls.git
+  TESTS_REPO: ssh://git@mybitbucket.llnl.gov:7999/mfem/tests.git
+  AUTOTEST_REPO: ssh://git@mybitbucket.llnl.gov:7999/mfem/autotest.git
+  MFEM_DATA_REPO: https://github.com/mfem/data.git
+  ARTIFACTS_DIR: artifacts
+  SLURM_OVERLAP: 1

 # The pipeline is divided into stages. Usually, jobs in a given stage wait for
 # the preceding stages to complete before to start. However, we sometimes use
 # the "needs" keyword and express the DAG of jobs for more efficiency.
 # - We use setup and setup_baseline phases to download content outside of mfem
 #   directory.
-# - Allocate/Release is where quartz resource are allocated/released once for all.
+# - Allocate/Release is where quartz resources are allocated/released once for all.
 # - Build and Test is where we build and MFEM for multiple toolchains.
 # - Baseline_checks gathers baseline-type test suites execution
 # - Baseline_publish, only available on master, allows to update baseline
 #   results
 stages:
-  - sub-pipelines
+  - setup
+  - q_allocate_resources
+  - q_build_and_test
+  - q_release_resources
+  - l_build_and_test
+  - c_build_and_test
+  - setup_baseline
+  - baseline_check
+  - baseline_to_autotest
+  - baseline_publish

-variables:
-  CUSTOM_CI_BUILDS_DIR: "/usr/workspace/mfem/gitlab-runner"
-  USER_CI_TOP_DIR: "${CUSTOM_CI_BUILDS_DIR}/${GITLAB_USER_LOGIN}"
-  SHARED_REPOS_DIR: "${USER_CI_TOP_DIR}/repos"
-  AUTOTEST_ROOT: "${SHARED_REPOS_DIR}"
-  # MFEM_DATA_DIR is setup in '.gitlab/configs/setup-build-and-test.yml' and
-  # used in '.gitlab/configs/<machine>-config.yml':
-  MFEM_DATA_DIR: "${SHARED_REPOS_DIR}/mfem-data"
-
-# Defines the default choice for updating the saved baseline results. By default
-# the baseline can only be updated from the master branch. This variable offers
-# the option to manually ask for rebaselining from another branch if necessary.
-  REBASELINE: "NO"
-  AUTOTEST: "NO"
-  # AUTOTEST_COMMIT: used only when AUTOTEST is set to YES.
-  # * If AUTOTEST_COMMIT is NOT set to NO, reporting jobs will commit their
-  #   files to the MFEM/autotest repo.
-  # * If AUTOTEST_COMMIT is set to NO, reporting jobs will NOT commit their
-  #   files to the MFEM/autotest repo. Instead they will just show the contents
-  #   of the report files and remove them.
-  AUTOTEST_COMMIT: "YES"
-
-# Trigger subpipelines:
-quartz-build-and-test:
-  stage: sub-pipelines
+# setup clones the mfem/data repo in ${BUILD_ROOT}. The build_and_test script
+# then symlinks the repo to the parent directory of the MFEM source directory.
+# Unit tests that depend on the mfem/data repo will then detect that this
+# directory is present and be enabled.
+setup:
+  tags:
+    - shell
+    - quartz
+  stage: setup
  variables:
-    # Explicitly pass down values that we want to be able to set when triggering
-    # pipelines manually or using scheduling
-    AUTOTEST: "${AUTOTEST}"
-    AUTOTEST_COMMIT: "${AUTOTEST_COMMIT}"
-  trigger:
-    include: .gitlab/quartz-build-and-test.yml
-    strategy: depend
+    GIT_STRATEGY: none
+  script:
+    - mkdir -p ${BUILD_ROOT} && cd ${BUILD_ROOT}
+    - if [ ! -d data ]; then git clone ${MFEM_DATA_REPO}; fi

-quartz-baseline:
-  stage: sub-pipelines
+# The setup_baseline job in setup stage_baseline doesn't rely on MFEM git repo.
+# It prepares a pipeline-wide working directory downloading/updating external
+# repos. TODO: updating tests and tpls is not necessary anymore since pipelines
+# are now using unique directories so repo are never shared with another
+# pipeline. This is not memory efficient (we keep a lot of data), hence this
+# reminder.
+# Note: This job can start immediately.
+setup_baseline:
+  tags:
+    - shell
+    - quartz
+  stage: setup_baseline
  variables:
-    # Explicitly pass down values that we want to be able to set when triggering
-    # pipelines manually or using scheduling
-    REBASELINE: "${REBASELINE}"
-    AUTOTEST: "${AUTOTEST}"
-    AUTOTEST_COMMIT: "${AUTOTEST_COMMIT}"
-  trigger:
-    include: .gitlab/quartz-baseline.yml
-    strategy: depend
+    GIT_STRATEGY: none
+  script:
+    - mkdir -p ${BUILD_ROOT} && cd ${BUILD_ROOT}
+    - if [ ! -d "tpls" ]; then git clone ${TPLS_REPO}; fi
+    - if [ ! -d "tests" ]; then git clone ${TESTS_REPO}; fi
+    - cd tpls && git pull && cd ..
+    - cd tests && git pull && cd ..
+    - cd ${AUTOTEST_ROOT}
+    - if [ ! -d "autotest" ]; then git clone ${AUTOTEST_REPO}; fi
+    - cd autotest && git pull && cd ..
+  needs: []

-lassen-build-and-test:
-  stage: sub-pipelines
-  variables:
-    # Explicitly pass down values that we want to be able to set when triggering
-    # pipelines manually or using scheduling
-    AUTOTEST: "${AUTOTEST}"
-    AUTOTEST_COMMIT: "${AUTOTEST_COMMIT}"
-  trigger:
-    include: .gitlab/lassen-build-and-test.yml
-    strategy: depend
+.build_toss_3_x86_64_ib_script:
+  script:
+    - export THREADS=12
+    - echo ${ALLOC_NAME}
+    - export JOBID=$(squeue -h --name=${ALLOC_NAME} --format=%A)
+    - echo ${JOBID}
+    - srun $( [[ -n "${JOBID}" ]] && echo "--jobid=${JOBID}" ) -t 30 -N 1 tests/gitlab/build_and_test

-corona-build-and-test:
-  stage: sub-pipelines
+.build_toss_3_x86_64_ib_corona_script:
+  script:
+    - srun -p mi60 -t 15 -N 1 tests/gitlab/build_and_test
+
+# Lassen uses a different job scheduler (spectrum lsf) that does not allow
+# pre-allocation the same way slurm does.  We use pdebug queue on lassen to
+# speed-up the allocation.  However this would not be scalable to multiple
+# builds.
+.build_blueos_3_ppc64le_ib_script:
+  script:
+    - lalloc 1 -W 30 -q pdebug tests/gitlab/build_and_test
+
+# Shared script for baseline and sample-run-baseline, the value of BASELINE_TEST
+# differentiates between the two tests.
+.baseline_script: &baseline_script |
+  # locals
+  _glob_err=${BASELINE_TEST}.err
+  _base_diff=${BASELINE_TEST}-${SYS_TYPE}.diff
+  _base_patch=${BASELINE_TEST}-${SYS_TYPE}.patch
+  _base_out=${BASELINE_TEST}-${SYS_TYPE}.out
+  # prepare
+  cd ${BUILD_ROOT}
+  ln -snf ${CI_PROJECT_DIR} mfem
+  cd tests
+  mkdir _${BASELINE_TEST} && cd _${BASELINE_TEST}
+  # run
+  srun --nodes=1 -p pdebug ../runtest ../../mfem "${BASELINE_TEST} ${ADDITIONAL_DIR}"
+  # post
+  mkdir ${CI_PROJECT_DIR}/${ARTIFACTS_DIR}
+  if [[ -s ${_glob_err} ]]
+  then
+    echo "ERROR during ${BASELINE_TEST} execution";
+    echo "Here is the ${_glob_err} file content";
+    cat ${_glob_err}
+    cp ${_glob_err} ${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/${_glob_err}
+    exit 1;
+  elif [[ ! -f ${_base_patch} && ! -f ${_base_out} ]]
+  then
+    echo "Something went WRONG in ${BASELINE_TEST}:";
+    echo "Either ${_base_patch} or ${_base_out} should exists";
+    exit 1;
+  elif [[ -f ${_base_patch} ]]
+  then
+    echo "${BASELINE_TEST}: Differences found, patch generated"
+    cp ${_base_patch} ${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/${_base_patch}
+  elif [[ -f ${_base_out} ]]
+  then
+    echo "${BASELINE_TEST}: Differences found, replacement file generated"
+    cp ${_base_out} ${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/${_base_out}
+  fi
+  # _base_diff won't even exist if there is no difference.
+  if [[ -f ${_base_diff} ]]
+  then
+    echo "${BASELINE_TEST}: Relevant differences (filtered diff) ..."
+    cat ${_base_diff}
+    cp ${_base_diff} ${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/${_base_diff}
+    # We create a .err file, because that's how we signal that there was a diff.
+    cp ${_base_diff} ${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/gitlab-${BASELINE_TEST}-${SYS_TYPE}.err
+  fi
+  if [[ ! -s ${_base_diff} ]]
+  then
+    echo "${BASELINE_TEST}: PASSED"
+    true
+  else
+    echo "${BASELINE_TEST}: FAILED"
+    false
+  fi
+
+# Actual templates for baseline checks
+.baselinecheck_mfem:
+  stage: baseline_check
  variables:
-    # Explicitly pass down values that we want to be able to set when triggering
-    # pipelines manually or using scheduling
-    AUTOTEST: "${AUTOTEST}"
-    AUTOTEST_COMMIT: "${AUTOTEST_COMMIT}"
-  trigger:
-    include: .gitlab/corona-build-and-test.yml
-    strategy: depend
+    BASELINE_TEST: baseline
+    ADDITIONAL_DIR: ${BUILD_ROOT}/tpls
+  script:
+    - *baseline_script
+  artifacts:
+    when: always
+    paths:
+      - ${ARTIFACTS_DIR}
+  allow_failure: true
+
+.samplebaselinecheck_mfem:
+  stage: baseline_check
+  variables:
+    BASELINE_TEST: sample-runs-baseline
+    ADDITIONAL_DIR: ""
+  script:
+    - *baseline_script
+  timeout: 4h
+  artifacts:
+    when: always
+    paths:
+      - ${ARTIFACTS_DIR}
+  allow_failure: true
+
+# This job can only be manually triggered on a pipeline for master branch, or if
+# the pipeline was triggered with REBASELINE="YES"
+.rebaseline_mfem:
+  stage: baseline_publish
+  rules:
+    - if: '$CI_COMMIT_BRANCH == "master" || $REBASELINE == "YES"'
+      when: manual
+  script:
+    - export PATCH_FILE=${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/baseline-${SYS_TYPE}.patch
+    - export FULL_FILE=${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/baseline-${SYS_TYPE}.out
+    - export DIFF_FILE=${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/baseline-${SYS_TYPE}.diff
+    - cd ${BUILD_ROOT}/tests
+    - |
+      if [[ ! -f "${DIFF_FILE}" ]]
+      then
+          echo "Nothing to be done: no relevant change in baseline"
+          exit 0
+      elif [[ -f "${PATCH_FILE}" ]]
+      then
+          patch "./baseline-${SYS_TYPE}.saved" < "${PATCH_FILE}"
+      elif [[ -f "${FULL_FILE}t" ]]
+      then
+          cp "${FULL_FILE}" "./baseline-${SYS_TYPE}.saved"
+      else
+          echo "File missing: expected ${PATCH_FILE} or ${FULL_FILE}"
+          exit 1
+      fi
+    - git add baseline-${SYS_TYPE}.saved
+    - git commit -m "${SYS_TYPE} rebaselined in GitLab pipeline ${CI_PIPELINE_ID}"
+    - git push origin master
+
+# The list on jobs is defined in machine-specific files.
+include:
+  - local: .gitlab/quartz.yml
+  - local: .gitlab/lassen.yml
@@ -1,94 +0,0 @@
-                    Finite Element Discretization Library
-                                   __
-                       _ __ ___   / _|  ___  _ __ ___
-                      | '_ ` _ \ | |_  / _ \| '_ ` _ \
-                      | | | | | ||  _||  __/| | | | | |
-                      |_| |_| |_||_|   \___||_| |_| |_|
-
-                               https://mfem.org
-
-
-This directory contains most of the GitLab CI configuration. MFEM runs both PR
-and nightly testing on GitLab.
-
-# Structure
-
-## Top level
-
-The root configuration file is `.gitlab-ci.yml` at the root of MFEM repo.
-This file only defines one stage, in which we trigger several
-sub-pipelines.
-
-We use sub-pipelines to isolate the test for one combination of `machine`
-and `test type`.
-
-Machines typically include:
-
-* Quartz: Intel bi-socket x86
-* Lassen: Power9 + Nvidia GPU
-* Corona: AMD GPU
-
-Test types include:
-
-* Build and test: Spack driven build of dependencies, mfem build, mfem
-  test
-* Baseline: Script driven build of dependencies, thorough testing
-
-⚠️ The sub-pipeline design allows to add a new machine or a new test type without
-altering the scheduling, execution and displaying of the others.
-
-## Sub-pipelines
-
-Each file is this directory is the root configuration file for one
-sub-pipeline. The naming reflects the corresponding couple (`machine`,
-`test_type`).
-
-Those files define the *stages* and the *jobs* for the sub-pipeline. They
-also contain any configuration that cannot be shared. For the most part
-though, the configuration is shared and is placed in `.gitlab/configs`.
-
-We try to keep scripts out of the CI config and share them among similar
-jobs. They are gathered in `.gitlab/scripts`.
-
-## Scripts
-
-Scripts specific to the CI only are in `.gitlab/scripts`. It is best practice
-to keep scripts outside the CI configuration (no bash scripts embedded in a
-yaml file) because it helps with readability, maintenance and also with
-transition to another CI system.
-
-⚠️ Most of the scripts there are driven by environment variables and do not have a
-usage function. This should be improved.
-
-
-# More testing
-
-## Adding a new target to a build_and_test pipeline
-
-`build_and_test` pipelines rely on Spack to install dependencies. Spack is
-driven by Uberenv which helps freezing Spack configuration: the goal being to
-point to specific commit in Spack and isolate its configuration so that it is
-not influenced by the user environment. More documentation about this can be
-found in `tests/gitlab`.
-
-In the end, the MFEM target for which to build the dependencies is expressed
-with a spack spec of MFEM, within the limits permitted by the MFEM spack
-package.
-
-In any build-and-test sub-pipeline a job basically consists in defining the
-spack spec to use. Adding a job on quartz for example resumes to:
-
-```yaml
-<job_name>:
-  variables:
-    SPEC: "<spack_spec>"
-  extends: .build_and_test_on_quartz
-```
-
-The remaining and non trivial work is to make sure this spec is working. To
-test a spec before adding it, or reproduce a CI configuration, please refer to
-`tests/gitlab/reproduce-ci-jobs-interactively.md`.
-
-⚠️ It is assumed that the spack spec applies to `mfem@develop`. That's why in the
-CI all the specs start with the compiler or the variants to apply to mfem. The
-mechanism still works with a full spec.
@@ -1,36 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# We define the following GitLab pipeline variables:
-variables:
-
-# The path to the shared resource between all jobs. For example, external
-# repositories like 'tests' and 'tpls' are cloned here. Also, 'tpls' is built
-# once for all targets, so that build happen here. The BUILD_ROOT is unique to
-# the pipeline, preventing any form of concurrency with other pipelines. This
-# also means that the BUILD_ROOT directory will never be cleaned.
-# TODO: add a clean-up mechanism
-  BUILD_ROOT: ${USER_CI_TOP_DIR}/${CI_PROJECT_NAME}-${MACHINE_NAME}-pipeline-${CI_PIPELINE_ID}
-
-# On LLNL's quartz, there is only one allocation shared among jobs in order to
-# save time and resource. This allocation has to be uniquely named so that we
-# are sure to retrieve it.
-  ALLOC_NAME: ${CI_PROJECT_NAME}_ci_${CI_PIPELINE_ID}
-
-# Git repositories used in the pipeline
-  TPLS_REPO: ssh://git@mybitbucket.llnl.gov:7999/mfem/tpls.git
-  TESTS_REPO: ssh://git@mybitbucket.llnl.gov:7999/mfem/tests.git
-  AUTOTEST_REPO: ssh://git@mybitbucket.llnl.gov:7999/mfem/autotest.git
-  MFEM_DATA_REPO: https://github.com/mfem/data.git
-
-# Directory used to place artifacts.
-  ARTIFACTS_DIR: artifacts
-  SLURM_OVERLAP: 1
@@ -1,59 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# GitLab pipeline configuration for the Corona machine at LLNL
-variables:
-  MACHINE_NAME: corona
-
-.on_corona:
-  tags:
-    - shell
-    - corona
-  rules:
-    # Don’t run corona jobs if...
-    # Note: This makes corona an "opt-in" machine. To activate builds on corona
-    # for a given GitLab clone of MFEM, go to Setting/CI-CD/variables, and set
-    # "ON_CORONA" to "ON". An LC account on for corona is required to trigger a
-    # pipeline there.
-    - if: '$CI_COMMIT_BRANCH =~ /_cnone/ || $ON_CORONA != "ON"'
-      when: never
-    # Don’t run autotest update if...
-    - if: '$CI_JOB_NAME =~ /report/ && $AUTOTEST != "YES"'
-      when: never
-    # Report success on success status
-    - if: '$CI_JOB_NAME =~ /report_job_success/ && $AUTOTEST == "YES"'
-      when: on_success
-    # Report failure on failure status
-    - if: '$CI_JOB_NAME =~ /report_job_failure/ && $AUTOTEST == "YES"'
-      when: on_failure
-    # Always release resource
-    - if: '$CI_JOB_NAME =~ /release_resource/'
-      when: always
-    # Always cleanup
-    - if: '$CI_JOB_NAME =~ /cleanup/'
-      when: always
-    # Default is to run if previous stage succeeded
-    - when: on_success
-
-# Spack helped builds
-# Generic corona build job, extending build script
-.build_and_test_on_corona:
-  extends: [.on_corona]
-  stage: build_and_test
-  script:
-    # THREADS is used by 'tests/gitlab/build_and_test', run below
-    - export THREADS=12
-    - echo ${ALLOC_NAME}
-    - export JOBID=$(squeue -h --name=${ALLOC_NAME} --format=%A)
-    - echo ${JOBID}
-    - echo ${MFEM_DATA_DIR}
-    - echo ${SPEC}
-    - srun $( [[ -n "${JOBID}" ]] && echo "--jobid=${JOBID}" ) -t 15 -N 1 tests/gitlab/build_and_test --spec "${SPEC}" --data-dir "${MFEM_DATA_DIR}" --data
@@ -1,49 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# GitLab pipelines configurations for the Lassen machine at LLNL
-variables:
-  MACHINE_NAME: lassen
-
-.on_lassen:
-  tags:
-    - shell
-    - lassen
-  rules:
-    - if: '$CI_COMMIT_BRANCH =~ /_lnone/ || $ON_LASSEN == "OFF"' #run except if ...
-      when: never
-    # Don't run autotest update if...
-    - if: '$CI_JOB_NAME =~ /report/ && $AUTOTEST != "YES"'
-      when: never
-    # Report success on success status
-    - if: '$CI_JOB_NAME =~ /report_job_success/ && $AUTOTEST == "YES"'
-      when: on_success
-    # Report failure on failure status
-    - if: '$CI_JOB_NAME =~ /report_job_failure/ && $AUTOTEST == "YES"'
-      when: on_failure
-    # Always cleanup
-    - if: '$CI_JOB_NAME =~ /cleanup/'
-      when: always
-    - when: on_success
-
-# Lassen uses a different job scheduler (spectrum lsf) that does not allow
-# pre-allocation the same way slurm does.  We use pdebug queue on lassen
-# to speed-up the allocation.  However this would not be scalable to
-# multiple builds.
-.build_and_test_on_lassen:
-  extends: [.on_lassen]
-  stage: build_and_test
-  script:
-    - echo ${MFEM_DATA_DIR}
-    - echo ${SPEC}
-    # Next script uses 'THREADS': leaving it empty --> it uses 'make all -j'
-    - lalloc 1 -W 30 -q pdebug --atsdisable tests/gitlab/build_and_test --spec "${SPEC}" --data-dir "${MFEM_DATA_DIR}" --data
-  needs: [setup]
@@ -1,55 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# GitLab pipelines configurations for the Quartz machine at LLNL
-variables:
-  MACHINE_NAME: quartz
-
-.on_quartz:
-  tags:
-    - shell
-    - quartz
-  rules:
-    # Don't run quartz jobs if...
-    - if: '$CI_COMMIT_BRANCH =~ /_qnone/ || $ON_QUARTZ == "OFF"'
-      when: never
-    # Don't run autotest update if...
-    - if: '$CI_JOB_NAME =~ /report/ && $AUTOTEST != "YES"'
-      when: never
-    # Report success on success status
-    - if: '$CI_JOB_NAME =~ /report_job_success/ && $AUTOTEST == "YES"'
-      when: on_success
-    # Report failure on failure status
-    - if: '$CI_JOB_NAME =~ /report_job_failure/ && $AUTOTEST == "YES"'
-      when: on_failure
-    # Always release resource
-    - if: '$CI_JOB_NAME =~ /release_resource/'
-      when: always
-    # Always cleanup
-    - if: '$CI_JOB_NAME =~ /cleanup/'
-      when: always
-    # Default is to run if previous stage succeeded
-    - when: on_success
-
-# Spack helped builds
-# Generic quartz build job, extending build script
-.build_and_test_on_quartz:
-  extends: [.on_quartz]
-  stage: build_and_test
-  script:
-    # THREADS is used by 'tests/gitlab/build_and_test', run below
-    - export THREADS=12
-    - echo ${ALLOC_NAME}
-    - export JOBID=$(squeue -h --name=${ALLOC_NAME} --format=%A)
-    - echo ${JOBID}
-    - echo ${MFEM_DATA_DIR}
-    - echo ${SPEC}
-    - srun $( [[ -n "${JOBID}" ]] && echo "--jobid=${JOBID}" ) -t 30 -N 1 tests/gitlab/build_and_test --spec "${SPEC}" --data-dir "${MFEM_DATA_DIR}" --data
@@ -1,81 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# Jobs report
-.report_job_success:
-  script:
-    - echo ${MACHINE_NAME}
-    - echo ${AUTOTEST}
-    - echo ${AUTOTEST_COMMIT}
-    - echo "AUTOTEST_ROOT ${AUTOTEST_ROOT}"
-    - cd ${AUTOTEST_ROOT}
-    - |
-      (
-        date
-        echo "Waiting to aquire lock on '$PWD/autotest.lock' ..."
-        # try to get an excusive lock on fd 9 (autotest.lock) repeating the try
-        # every 5 seconds; simply using no timeout, i.e. 'flock 9', causes the
-        # command to hang indefinitely sometimes, so we use the timeout & retry
-        # as a workaround; we may want to add a counter for the number of
-        # retries to interrupt a potential infinite loop
-        while ! flock -w 5 9; do
-          true
-        done
-        echo "Aquired lock on '$PWD/autotest.lock'"
-        date
-        # Report SUCCESS while holding the file lock on 'autotest.lock'.
-        # The next script uses the following environment variables:
-        # - MACHINE_NAME, AUTOTEST_ROOT, AUTOTEST_COMMIT
-        # - CI_COMMIT_REF_SLUG, CI_PROJECT_DIR, CI_PIPELINE_URL
-        # It also calls the script '.gitlab/scripts/safe_create_rundir'.
-        ${CI_PROJECT_DIR}/.gitlab/scripts/report_build_and_test_success
-        err=$?
-        # sleep for a period to allow NFS to propagate the above changes;
-        # clearly, there is no guarantee that other NFS clients will see the
-        # changes even after the timeout
-        sleep 10
-        exit $err
-      ) 9> autotest.lock
-
-.report_job_failure:
-  script:
-    - echo ${MACHINE_NAME}
-    - echo ${AUTOTEST}
-    - echo ${AUTOTEST_COMMIT}
-    - echo "AUTOTEST_ROOT ${AUTOTEST_ROOT}"
-    - cd ${AUTOTEST_ROOT}
-    - |
-      (
-        date
-        echo "Waiting to aquire lock on '$PWD/autotest.lock' ..."
-        # try to get an excusive lock on fd 9 (autotest.lock) repeating the try
-        # every 5 seconds; simply using no timeout, i.e. 'flock 9', causes the
-        # command to hang indefinitely sometimes, so we use the timeout & retry
-        # as a workaround; we may want to add a counter for the number of
-        # retries to interrupt a potential infinite loop
-        while ! flock -w 5 9; do
-          true
-        done
-        echo "Aquired lock on '$PWD/autotest.lock'"
-        date
-        # Report FAILURE while holding the file lock on 'autotest.lock'.
-        # The next script uses the following environment variables:
-        # - MACHINE_NAME, AUTOTEST_ROOT, AUTOTEST_COMMIT
-        # - CI_COMMIT_REF_SLUG, CI_PROJECT_DIR, CI_PIPELINE_URL
-        # It also calls the script '.gitlab/scripts/safe_create_rundir'.
-        ${CI_PROJECT_DIR}/.gitlab/scripts/report_build_and_test_failure
-        err=$?
-        # sleep for a period to allow NFS to propagate the above changes;
-        # clearly, there is no guarantee that other NFS clients will see the
-        # changes even after the timeout
-        sleep 10
-        exit $err
-      ) 9> autotest.lock
@@ -1,72 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# The setup_baseline job doesn't rely on MFEM git repo. It prepares a
-# pipeline-wide working directory downloading/updating external repos.
-# TODO:
-# updating tests and tpls is not necessary anymore since pipelines are now
-# using unique directories so repo are never shared with another pipeline. This
-# is not memory efficient (we keep a lot of data), hence this reminder.
-setup_baseline:
-  tags:
-    - shell
-    - quartz
-  stage: setup
-  variables:
-    GIT_STRATEGY: none
-  script:
-    #
-    # Setup ${BUILD_ROOT}/tpls and ${BUILD_ROOT}/tests:
-    #
-    - echo "MACHINE_NAME = ${MACHINE_NAME}"
-    - echo "REBASELINE = ${REBASELINE}"
-    - echo "AUTOTEST = ${AUTOTEST}"
-    - echo "AUTOTEST_COMMIT = ${AUTOTEST_COMMIT}"
-    - echo "BUILD_ROOT ${BUILD_ROOT}"
-    - mkdir -p ${BUILD_ROOT} && cd ${BUILD_ROOT}
-    - if [ ! -d "tpls" ]; then git clone ${TPLS_REPO}; fi
-    - if [ ! -d "tests" ]; then git clone ${TESTS_REPO}; fi
-    - cd tpls && git pull && cd ..
-    - cd tests && git pull origin && cd ..
-    #
-    # Setup ${AUTOTEST_ROOT}/autotest:
-    #
-    - echo "AUTOTEST_ROOT ${AUTOTEST_ROOT}"
-    - mkdir -p ${AUTOTEST_ROOT} && cd ${AUTOTEST_ROOT}
-    - command -v flock || echo "Required command 'flock' not found"
-    - |
-      (
-        date
-        echo "Waiting to aquire lock on '$PWD/autotest.lock' ..."
-        # try to get an excusive lock on fd 9 (autotest.lock) repeating the try
-        # every 5 seconds; simply using no timeout, i.e. 'flock 9', causes the
-        # command to hang indefinitely sometimes, so we use the timeout & retry
-        # as a workaround; we may want to add a counter for the number of
-        # retries to interrupt a potential infinite loop
-        while ! flock -w 5 9; do
-          true
-        done
-        echo "Aquired lock on '$PWD/autotest.lock'"
-        date
-        # clone/update the autotest repo while holding the file lock on
-        # 'autotest.lock'
-        err=0
-        if [[ ! -d "autotest" ]]; then
-          git clone ${AUTOTEST_REPO}
-        else
-          cd autotest && git pull && cd ..
-        fi || err=1
-        # sleep for a period to allow NFS to propagate the above changes;
-        # clearly, there is no guarantee that other NFS clients will see the
-        # changes even after the timeout
-        sleep 10
-        exit $err
-      ) 9> autotest.lock
@@ -1,94 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# Setup clones the mfem/data repo in ${SHARED_REPOS_DIR}. The build_and_test
-# script then symlinks the repo to the parent directory of the MFEM source
-# directory. Unit tests that depend on the mfem/data repo will then detect that
-# this directory is present and be enabled.
-setup:
-  tags:
-    - shell
-    - quartz
-  stage: setup
-  variables:
-    GIT_STRATEGY: none
-  script:
-    #
-    # Setup MFEM_DATA_DIR=${SHARED_REPOS_DIR}/mfem-data, see '.gitlab-ci.yml'
-    # and '.gitlab/configs/<machine>-config.yml'
-    #
-    - echo "MACHINE_NAME = ${MACHINE_NAME}"
-    - echo "AUTOTEST = ${AUTOTEST}"
-    - echo "AUTOTEST_COMMIT = ${AUTOTEST_COMMIT}"
-    - echo "SHARED_REPOS_DIR ${SHARED_REPOS_DIR}"
-    - mkdir -p ${SHARED_REPOS_DIR} && cd ${SHARED_REPOS_DIR}
-    - command -v flock || echo "Required command 'flock' not found"
-    - |
-      (
-        date
-        echo "Waiting to aquire lock on '$PWD/mfem-data.lock' ..."
-        # try to get an excusive lock on fd 9 (mfem-data.lock) repeating the try
-        # every 5 seconds; simply using no timeout, i.e. 'flock 9', causes the
-        # command to hang indefinitely sometimes, so we use the timeout & retry
-        # as a workaround; we may want to add a counter for the number of
-        # retries to interrupt a potential infinite loop
-        while ! flock -w 5 9; do
-          true
-        done
-        echo "Aquired lock on '$PWD/mfem-data.lock'"
-        date
-        # clone/update the mfem/data repo while holding the file lock on
-        # 'mfem-data.lock'
-        err=0
-        if [[ ! -d "mfem-data" ]]; then
-          git clone ${MFEM_DATA_REPO} "mfem-data"
-        else
-          cd "mfem-data" && git pull && cd ..
-        fi || err=1
-        # sleep for a period to allow NFS to propagate the above changes;
-        # clearly, there is no guarantee that other NFS clients will see the
-        # changes even after the timeout
-        sleep 10
-        exit $err
-      ) 9> mfem-data.lock
-    #
-    # Setup ${AUTOTEST_ROOT}/autotest:
-    #
-    - echo "AUTOTEST_ROOT ${AUTOTEST_ROOT}"
-    - mkdir -p ${AUTOTEST_ROOT} && cd ${AUTOTEST_ROOT}
-    - |
-      (
-        date
-        echo "Waiting to aquire lock on '$PWD/autotest.lock' ..."
-        # try to get an excusive lock on fd 9 (autotest.lock) repeating the try
-        # every 5 seconds; simply using no timeout, i.e. 'flock 9', causes the
-        # command to hang indefinitely sometimes, so we use the timeout & retry
-        # as a workaround; we may want to add a counter for the number of
-        # retries to interrupt a potential infinite loop
-        while ! flock -w 5 9; do
-          true
-        done
-        echo "Aquired lock on '$PWD/autotest.lock'"
-        date
-        # clone/update the autotest repo while holding the file lock on
-        # 'autotest.lock'
-        err=0
-        if [[ ! -d "autotest" ]]; then
-          git clone ${AUTOTEST_REPO}
-        else
-          cd autotest && git pull && cd ..
-        fi || err=1
-        # sleep for a period to allow NFS to propagate the above changes;
-        # clearly, there is no guarantee that other NFS clients will see the
-        # changes even after the timeout
-        sleep 10
-        exit $err
-      ) 9> autotest.lock
@@ -1,67 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-stages:
-  - setup
-  - allocate_resource
-  - build_and_test
-  - release_resource_and_report
-
-# Slurm shared allocation
-allocate_resource:
-  variables:
-    GIT_STRATEGY: none
-  extends: .on_corona
-  stage: allocate_resource
-  script:
-    - echo ${ALLOC_NAME}
-    - salloc --exclusive --nodes=1 --partition=mi60 --time=30 --no-shell --job-name=${ALLOC_NAME}
-  timeout: 6h
-  needs: [setup]
-
-# Build and test jobs, simply provide a spec
-rocm_gcc_8.3.1:
-  variables:
-    SPEC: "@develop%gcc@8.3.1+rocm amdgpu_target=gfx906"
-  extends: .build_and_test_on_corona
-  needs: [allocate_resource]
-
-# Release slurm allocation
-release_resource:
-  variables:
-    GIT_STRATEGY: none
-  extends: .on_corona
-  stage: release_resource_and_report
-  script:
-    - echo ${ALLOC_NAME}
-    - export JOBID=$(squeue -h --name=${ALLOC_NAME} --format=%A)
-    - echo ${JOBID}
-    - ([[ -n "${JOBID}" ]] && scancel ${JOBID})
-  needs: [rocm_gcc_8.3.1]
-
-# Jobs report
-report_job_success:
-  stage: release_resource_and_report
-  extends:
-    - .on_corona
-    - .report_job_success
-
-report_job_failure:
-  stage: release_resource_and_report
-  extends:
-    - .on_corona
-    - .report_job_failure
-
-include:
-  - local: .gitlab/configs/common.yml
-  - local: .gitlab/configs/corona-config.yml
-  - local: .gitlab/configs/setup-build-and-test.yml
-  - local: .gitlab/configs/report-build-and-test.yml
@@ -9,36 +9,26 @@
 # terms of the BSD-3 license. We welcome feedback and contributions, see file
 # CONTRIBUTING.md for details.

-stages:
-  - setup
-  - build_and_test
-  - report
+# GitLab pipelines configurations for the Lassen machine at LLNL
+
+.on_lassen:
+  tags:
+    - shell
+    - lassen
+  rules:
+    - if: '$CI_COMMIT_BRANCH =~ /_lnone/ || $ON_LASSEN == "OFF"' #run except if ...
+      when: never
+    - when: on_success
+
+# Spack helped builds
+# Generic lassen build job, extending build script
+# Note: Lassen jobs can start as soon as the setup job is complete.
+.build_and_test_on_lassen:
+  extends: [.build_blueos_3_ppc64le_ib_script, .on_lassen]
+  stage: l_build_and_test
+  needs: [setup]

 opt_mpi_cuda_xl_16_1_1_8:
  variables:
    SPEC: "%xl@16.1.1.8 +mpi +cuda cuda_arch=70"
  extends: .build_and_test_on_lassen
-
-opt_mpi_cuda_hypre_cuda_xl:
-  variables:
-    SPEC: "%xl@16.1.1.8 +mpi +cuda cuda_arch=70 ^hypre+cuda~shared cuda_arch=70"
-  extends: .build_and_test_on_lassen
-
-# Jobs report
-report_job_success:
-  stage: report
-  extends:
-    - .on_lassen
-    - .report_job_success
-
-report_job_failure:
-  stage: report
-  extends:
-    - .on_lassen
-    - .report_job_failure
-
-include:
-  - local: .gitlab/configs/common.yml
-  - local: .gitlab/configs/lassen-config.yml
-  - local: .gitlab/configs/setup-build-and-test.yml
-  - local: .gitlab/configs/report-build-and-test.yml
@@ -1,133 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-variables:
-  BASELINE_TEST: baseline
-
-stages:
-  - setup
-  - baseline_check
-  - baseline_report
-  - cleanup
-  - baseline_publish
-
-baselinecheck_mfem_intel_quartz:
-  extends: [.on_quartz]
-  stage: baseline_check
-  variables:
-    # TPLS_DIR is used in .gitlab/scripts/baseline to provide the tpls location
-    # when call the runtest script in MFEM test repo.
-    # Note: the value must be consistent with the setup performed in
-    #       .gitlab/configs/setup-baseline.yml.
-    TPLS_DIR: ${BUILD_ROOT}/tpls
-  script:
-    - echo ${BUILD_ROOT}
-    - echo ${TPLS_DIR}
-    # Used by the tests in MFEM/tests:
-    - export MFEM_TEST_NP=32
-    # The next script uses the following environment variables:
-    # * BASELINE_TEST, SYS_TYPE, CI_PROJECT_DIR, ARTIFACTS_DIR,
-    # * BUILD_ROOT, TPLS_DIR, MACHINE_NAME
-    - .gitlab/scripts/baseline
-  artifacts:
-    when: always
-    paths:
-      - ${ARTIFACTS_DIR}
-  allow_failure: true
-
-cleanup:
-  extends: .on_quartz
-  stage: cleanup
-  variables:
-    GIT_STRATEGY: none
-  script:
-    - echo "BUILD_ROOT=${BUILD_ROOT}"
-    - rm -rf "${BUILD_ROOT}" || true
-
-report_baseline:
-  extends: [.on_quartz]
-  stage: baseline_report
-  script:
-    - echo ${MACHINE_NAME}
-    - echo ${AUTOTEST}
-    - echo ${AUTOTEST_COMMIT}
-    - echo "AUTOTEST_ROOT ${AUTOTEST_ROOT}"
-    - cd ${AUTOTEST_ROOT}
-    - |
-      (
-        date
-        echo "Waiting to aquire lock on '$PWD/autotest.lock' ..."
-        # try to get an excusive lock on fd 9 (autotest.lock) repeating the try
-        # every 5 seconds; simply using no timeout, i.e. 'flock 9', causes the
-        # command to hang indefinitely sometimes, so we use the timeout & retry
-        # as a workaround; we may want to add a counter for the number of
-        # retries to interrupt a potential infinite loop
-        while ! flock -w 5 9; do
-          true
-        done
-        echo "Aquired lock on '$PWD/autotest.lock'"
-        date
-        # ----------------------
-        cd ${AUTOTEST_ROOT}/autotest || \
-          { echo "Invalid 'autotest' dir: ${AUTOTEST_ROOT}/autotest"; exit 1; }
-        mkdir -p ${MACHINE_NAME}
-        rundir="${MACHINE_NAME}/$(date +%Y-%m-%d)-gitlab-${BASELINE_TEST}-${CI_COMMIT_REF_SLUG}"
-        rundir=$(${CI_PROJECT_DIR}/.gitlab/scripts/safe_create_rundir ${rundir})
-        cp ${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/* ${rundir}
-        # We create an autotest-email.html file, because that's how we signal that there was a diff (temporary).
-        if [[ -f ${rundir}/${BASELINE_TEST}.err ]]; then
-          cp ${rundir}/${BASELINE_TEST}.err ${rundir}/autotest-email.html
-        fi
-        printf "%s\n" "" "Pipeline URL:" "$CI_PIPELINE_URL" \
-          >> ${rundir}/pipeline.txt
-        msg="GitLab CI log for ${BASELINE_TEST} on ${MACHINE_NAME} ($(date +%Y-%m-%d))"
-        if [[ "$AUTOTEST_COMMIT" != "NO" ]]; then
-          git pull && \
-          git add ${rundir} && \
-          git commit -m "${msg}" && \
-          git push origin master
-        else
-          for file in ${rundir}/*; do
-            echo "------------------------------"
-            echo "Content of '$file'"
-            echo "******************************"
-            cat $file
-            echo "******************************"
-          done
-          rm -rf ${rundir} || true
-        fi
-        err=$?
-        # ----------------------
-        # sleep for a period to allow NFS to propagate the above changes;
-        # clearly, there is no guarantee that other NFS clients will see the
-        # changes even after the timeout
-        sleep 10
-        exit $err
-      ) 9> autotest.lock
-
-baselinepublish_mfem_quartz:
-  extends: [.on_quartz]
-  stage: baseline_publish
-  rules:
-    # - if: '$CI_COMMIT_BRANCH == "master" || $REBASELINE == "YES"'
-    - if: '$REBASELINE == "YES"'
-      when: manual
-  script:
-    - echo ${BUILD_ROOT}
-    - echo ${PWD}
-    - echo ${ARTIFACTS_DIR}
-    - ls -lA ${ARTIFACTS_DIR}
-    - .gitlab/scripts/rebaseline
-
-include:
-  - local: .gitlab/configs/common.yml
-  - local: .gitlab/configs/quartz-config.yml
-  - local: .gitlab/configs/setup-baseline.yml
@@ -1,99 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-stages:
-  - setup
-  - allocate_resource
-  - build_and_test
-  - release_resource_and_report
-
-# Allocate
-allocate_resource:
-  variables:
-    GIT_STRATEGY: none
-  extends: .on_quartz
-  stage: allocate_resource
-  script:
-    - echo ${ALLOC_NAME}
-    - salloc --exclusive --nodes=1 --partition=pdebug --time=30 --no-shell --job-name=${ALLOC_NAME}
-  timeout: 6h
-
-# GitLab jobs for the Quartz machine at LLNL
-debug_ser_gcc_4_9_3:
-  variables:
-    SPEC: "%gcc@4.9.3 +debug~mpi"
-  extends: .build_and_test_on_quartz
-
-debug_ser_gcc_6_1_0:
-  variables:
-    SPEC: "%gcc@6.1.0 +debug~mpi"
-  extends: .build_and_test_on_quartz
-
-debug_par_gcc_6_1_0:
-  variables:
-    SPEC: "%gcc@6.1.0 +debug+mpi"
-  extends: .build_and_test_on_quartz
-
-opt_ser_gcc_6_1_0:
-  variables:
-    SPEC: "%gcc@6.1.0 ~mpi"
-  extends: .build_and_test_on_quartz
-
-opt_par_gcc_6_1_0:
-  variables:
-    SPEC: "%gcc@6.1.0"
-  extends: .build_and_test_on_quartz
-
-opt_par_gcc_6_1_0_sundials:
-  variables:
-    SPEC: "%gcc@6.1.0 +sundials"
-  extends: .build_and_test_on_quartz
-
-opt_par_gcc_6_1_0_petsc:
-  variables:
-    SPEC: "%gcc@6.1.0 +petsc ^petsc+mumps~superlu-dist"
-  extends: .build_and_test_on_quartz
-
-opt_par_gcc_6_1_0_pumi:
-  variables:
-    SPEC: "%gcc@6.1.0 +pumi"
-  extends: .build_and_test_on_quartz
-
-# Release
-release_resource:
-  variables:
-    GIT_STRATEGY: none
-  extends: .on_quartz
-  stage: release_resource_and_report
-  script:
-    - echo ${ALLOC_NAME}
-    - export JOBID=$(squeue -h --name=${ALLOC_NAME} --format=%A)
-    - echo ${JOBID}
-    - ([[ -n "${JOBID}" ]] && scancel ${JOBID})
-
-# Jobs report
-report_job_success:
-  stage: release_resource_and_report
-  extends:
-    - .on_quartz
-    - .report_job_success
-
-report_job_failure:
-  stage: release_resource_and_report
-  extends:
-    - .on_quartz
-    - .report_job_failure
-
-include:
-  - local: .gitlab/configs/common.yml
-  - local: .gitlab/configs/quartz-config.yml
-  - local: .gitlab/configs/setup-build-and-test.yml
-  - local: .gitlab/configs/report-build-and-test.yml
@@ -0,0 +1,184 @@
+# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
+# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
+# LICENSE and NOTICE for details. LLNL-CODE-806117.
+#
+# This file is part of the MFEM library. For more information and source code
+# availability visit https://mfem.org.
+#
+# MFEM is free software; you can redistribute it and/or modify it under the
+# terms of the BSD-3 license. We welcome feedback and contributions, see file
+# CONTRIBUTING.md for details.
+
+# GitLab pipelines configurations for the Quartz machine at LLNL
+
+.on_quartz:
+  tags:
+    - shell
+    - quartz
+  rules:
+    # Don't run quartz jobs if...
+    - if: '$CI_COMMIT_BRANCH =~ /_qnone/ || $ON_QUARTZ == "OFF"'
+      when: never
+    # Don't run autotest update if...
+    - if: '$CI_JOB_NAME =~ /update_autotest/ && $AUTOTEST != "YES"'
+      when: never
+    # Don't run autotest update if...
+    - if: '$CI_JOB_NAME =~ /q_report/ && $AUTOTEST != "YES"'
+      when: never
+    # Report success on success status
+    - if: '$CI_JOB_NAME =~ /q_report_success/ && $AUTOTEST == "YES"'
+      when: on_success
+    # Report failure on failure status
+    - if: '$CI_JOB_NAME =~ /q_report_failure/ && $AUTOTEST == "YES"'
+      when: on_failure
+    # Always release resources
+    - if: '$CI_JOB_NAME =~ /release_resources/'
+      when: always
+    # Default is to run if previous stage succeeded
+    - when: on_success
+
+# This is a yaml anchor, it can be used to avoid duplication like here.
+# The code below will simply be pasted wherever the anchor is placed.
+.safe_create_rundir: &safe_create_rundir |
+  if ! mkdir ${rundir}; then
+    n=1
+    while ! mkdir ${rundir}_${n}
+    do
+      n=$((n+1))
+    done
+    rundir=${rundir}_${n}
+  fi
+
+# Allocate
+q_allocate_resources:
+  variables:
+    GIT_STRATEGY: none
+  extends: .on_quartz
+  stage: q_allocate_resources
+  script:
+    - salloc --exclusive --nodes=1 --partition=pdebug --time=30 --no-shell --job-name=${ALLOC_NAME}
+  timeout: 6h
+
+# Release
+q_release_resources:
+  variables:
+    GIT_STRATEGY: none
+  extends: .on_quartz
+  stage: q_release_resources
+  script:
+    - export JOBID=$(squeue -h --name=${ALLOC_NAME} --format=%A)
+    - ([[ -n "${JOBID}" ]] && scancel ${JOBID})
+
+# Release
+q_report_success:
+  variables:
+    GIT_STRATEGY: none
+  extends: .on_quartz
+  stage: q_release_resources
+  script:
+    - echo "Can only run if all the quartz jobs passed"
+    - cd ${AUTOTEST_ROOT}/autotest && git pull
+    - rundir="gitlab/$(date +%Y-%m-%d)-github-${CI_COMMIT_REF_SLUG}"
+    - *safe_create_rundir
+    - echo "The Quartz jobs were successful" > ${rundir}/gitlab.out
+    - echo "See the pipeline here -> $CI_PIPELINE_URL" >> ${rundir}/gitlab.err
+    - git add ${rundir}
+    - git commit -am "Gitlab CI log for baseline on quartz with intel ($(date +%Y-%m-%d))"
+    - git push origin master
+
+q_report_failure:
+  variables:
+    GIT_STRATEGY: none
+  extends: .on_quartz
+  stage: q_release_resources
+  script:
+    - echo "Runs if there was at least one failure on quartz"
+    - cd ${AUTOTEST_ROOT}/autotest && git pull
+    - rundir="gitlab/$(date +%Y-%m-%d)-github-${CI_COMMIT_REF_SLUG}"
+    - *safe_create_rundir
+    - echo "There was an error while running CI on Quartz" > ${rundir}/gitlab.err
+    - echo "See the pipeline here -> $CI_PIPELINE_URL" >> ${rundir}/gitlab.err
+    - cp ${rundir}/gitlab.err ${rundir}/autotest-email.html
+    - git add ${rundir}
+    - git commit -am "Gitlab CI log for baseline on quartz with intel ($(date +%Y-%m-%d))"
+    - git push origin master
+
+# Spack helped builds
+# Generic quartz build job, extending build script
+.build_and_test_on_quartz:
+  extends: [.build_toss_3_x86_64_ib_script, .on_quartz]
+  stage: q_build_and_test
+
+# Build MFEM
+debug_ser_gcc_4_9_3:
+  variables:
+    SPEC: "%gcc@4.9.3 +debug~mpi"
+  extends: .build_and_test_on_quartz
+
+debug_ser_gcc_6_1_0:
+  variables:
+    SPEC: "%gcc@6.1.0 +debug~mpi"
+  extends: .build_and_test_on_quartz
+
+debug_par_gcc_6_1_0:
+  variables:
+    SPEC: "%gcc@6.1.0 +debug+mpi"
+  extends: .build_and_test_on_quartz
+
+opt_ser_gcc_6_1_0:
+  variables:
+    SPEC: "%gcc@6.1.0 ~mpi"
+  extends: .build_and_test_on_quartz
+
+opt_par_gcc_6_1_0:
+  variables:
+    SPEC: "%gcc@6.1.0"
+  extends: .build_and_test_on_quartz
+
+opt_par_gcc_6_1_0_sundials:
+  variables:
+    SPEC: "%gcc@6.1.0 +sundials"
+  extends: .build_and_test_on_quartz
+
+opt_par_gcc_6_1_0_petsc:
+  variables:
+    SPEC: "%gcc@6.1.0 +petsc ^petsc+mumps"
+  extends: .build_and_test_on_quartz
+
+opt_par_gcc_6_1_0_pumi:
+  variables:
+    SPEC: "%gcc@6.1.0 +pumi"
+  extends: .build_and_test_on_quartz
+
+# Baseline jobs form an independent set of jobs. We use `needs:[]` to specify
+# that "setup-baseline" can start immediately. Then, we have to use needs for
+# each one of the baseline jobs, otherwise they will wait for the rest of the
+# pipeline.
+
+baselinecheck_mfem_intel_quartz:
+  extends: [.baselinecheck_mfem, .on_quartz]
+  needs: [setup_baseline]
+
+update_autotest:
+  extends: [.on_quartz]
+  needs: [baselinecheck_mfem_intel_quartz]
+  stage: baseline_to_autotest
+  script:
+    - cd ${AUTOTEST_ROOT}/autotest && git pull
+    - rundir="quartz/$(date +%Y-%m-%d)-github-${CI_COMMIT_REF_SLUG}"
+    - *safe_create_rundir
+    - cp ${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/* ${rundir}
+    # We create an autotest-email.html file, because that's how we signal that there was a diff (temporary).
+    - |
+      if [[ -f ${rundir}/*.err ]]
+      then
+        echo "See the pipeline here -> $CI_PIPELINE_URL" >> ${rundir}/*.err
+        cp ${rundir}/*.err ${rundir}/autotest-email.html
+      fi
+    - git add ${rundir}
+    - git commit -am "Gitlab CI log for baseline on quartz with intel ($(date +%Y-%m-%d))"
+    - git push origin master
+
+baselinepublish_mfem_quartz:
+  extends: [.on_quartz, .rebaseline_mfem]
+  needs: [baselinecheck_mfem_intel_quartz]
@@ -1,88 +0,0 @@
-#!/bin/bash
-
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# locals
-glob_err=${BASELINE_TEST}.err
-base=${BASELINE_TEST}-${SYS_TYPE}
-base_diff=${base}.diff
-base_patch=${base}.patch
-base_out=${base}.out
-artifacts_path=${CI_PROJECT_DIR}/${ARTIFACTS_DIR}
-
-# prepare
-cd ${BUILD_ROOT} || \
-  { echo "Invalid BUILD_ROOT=$BUILD_ROOT"; exit 1; }
-ln -snf ${CI_PROJECT_DIR} mfem
-cd tests
-[[ -d _${BASELINE_TEST} ]] && rm -rf _${BASELINE_TEST}
-mkdir _${BASELINE_TEST} && cd _${BASELINE_TEST}
-
-# run
-if [[ "${MACHINE_NAME}" == "quartz" || "${MACHINE_NAME}" == "ruby" ]]; then
-  salloc --nodes=1 -p pdebug ../runtest ../../mfem "${BASELINE_TEST} ${TPLS_DIR}"
-elif [[ ${MACHINE_NAME} == "corona" ]]; then
-  salloc --nodes=1 -t 60 -p pbatch ../runtest ../../mfem "${BASELINE_TEST} ${TPLS_DIR}"
-elif [[ ${MACHINE_NAME} == "lassen" ]]; then
-  lalloc 1 -q pdebug ../runtest ../../mfem "${BASELINE_TEST} ${TPLS_DIR}"
-else
-  echo "Unknown machine: MACHINE_NAME=$MACHINE_NAME"
-  exit 1
-fi
-
-# post
-mkdir ${artifacts_path}
-
-if [[ -s ${glob_err} ]]
-then
-  echo "ERROR during ${BASELINE_TEST} execution";
-  echo "Here is the ${glob_err} file content";
-  cat ${glob_err}
-  cp ${glob_err} ${artifacts_path}/${glob_err}
-  exit 1;
-elif [[ ! -f ${base_patch} && ! -f ${base_out} ]]
-then
-  echo "Something went WRONG in ${BASELINE_TEST}:";
-  echo "Either ${base_patch} or ${base_out} should exists";
-  exit 1;
-elif [[ -f ${base_patch} ]]
-then
-  echo "${BASELINE_TEST}: Differences found, patch generated"
-  cp ${base_patch} ${artifacts_path}/${base_patch}
-elif [[ -f ${base_out} ]]
-then
-  echo "${BASELINE_TEST}: Differences found, replacement file generated"
-  cp ${base_out} ${artifacts_path}/${base_out}
-fi
-
-if [[ -f ${BASELINE_TEST}.out ]]; then
-  cp ${BASELINE_TEST}.out ${artifacts_path}
-fi
-
-# base_diff won't even exist if there is no difference.
-if [[ -f ${base_diff} ]]
-then
-  echo "${BASELINE_TEST}: Relevant differences (filtered diff) ..."
-  cat ${base_diff}
-  cp ${base_diff} ${artifacts_path}/${base_diff}
-  # We create a .err file, because that's how we signal that there was a diff.
-  cp ${base_diff} ${artifacts_path}/gitlab-${BASELINE_TEST}-${MACHINE_NAME}.err
-fi
-
-if [[ ! -s ${base_diff} ]]
-then
-  echo "${BASELINE_TEST}: PASSED"
-  true
-else
-  echo "${BASELINE_TEST}: FAILED"
-  false
-fi
@@ -1,49 +0,0 @@
-#!/bin/bash
-
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# There will be collision between corona and quartz baselines.
-# Once the corresponding files have been generated, we can switch to machine
-# specific ref.
-ARTIFACT_PATH=${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/baseline-${SYS_TYPE}
-#ARTIFACT_PATH=${CI_PROJECT_DIR}/${ARTIFACTS_DIR}/baseline-${SYS_TYPE}-${MACHINE_NAME}
-
-PATCH_FILE=${ARTIFACT_PATH}.patch
-FULL_FILE=${ARTIFACT_PATH}.out
-DIFF_FILE=${ARTIFACT_PATH}.diff
-
-# There will be collision between corona and quartz baselines.
-# Once the corresponding files have been generated, we can switch to machine
-# specific ref.
-SAVED_NAME=baseline-${SYS_TYPE}.saved
-#SAVED_NAME=baseline-${SYS_TYPE}-${MACHINE_NAME}.saved
-
-cd ${BUILD_ROOT}/tests
-
-if [[ ! -f "${DIFF_FILE}" ]]
-then
-    echo "Nothing to be done: no relevant change in baseline"
-    exit 0
-elif [[ -f "${PATCH_FILE}" ]]
-then
-    patch "${SAVED_NAME}" < "${PATCH_FILE}"
-elif [[ -f "${FULL_FILE}" ]]
-then
-    cp "${FULL_FILE}" "${SAVED_NAME}"
-else
-    echo "File missing: expected ${PATCH_FILE} or ${FULL_FILE}"
-    exit 1
-fi
-
-git add "${SAVED_NAME}"
-git commit -m "${SYS_TYPE} (${MACHINE_NAME}) rebaselined in GitLab pipeline ${CI_PIPELINE_ID}"
-git push origin master
@@ -1,45 +0,0 @@
-#!/bin/bash
-
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-echo "Runs if there was at least one failure on ${MACHINE_NAME}"
-
-cd ${AUTOTEST_ROOT}/autotest || \
-  { echo "Invalid 'autotest' dir: ${AUTOTEST_ROOT}/autotest"; exit 1; }
-mkdir -p ${MACHINE_NAME}
-
-rundir="${MACHINE_NAME}/$(date +%Y-%m-%d)-gitlab-ci-${CI_COMMIT_REF_SLUG}"
-rundir=$(${CI_PROJECT_DIR}/.gitlab/scripts/safe_create_rundir $rundir)
-
-printf "%s\n" "Some 'build-and-test' jobs on ${MACHINE_NAME} FAILED." \
-  "Pipeline URL:" "$CI_PIPELINE_URL" > ${rundir}/gitlab.err
-
-msg="GitLab CI log for build-and-test on ${MACHINE_NAME} ($(date +%Y-%m-%d))"
-
-# Create 'autotest-email.html' to indicate failure:
-cp ${rundir}/gitlab.err ${rundir}/autotest-email.html
-
-if [[ "$AUTOTEST_COMMIT" != "NO" ]]; then
-  git pull && \
-  git add ${rundir} && \
-  git commit -m "${msg}" && \
-  git push origin master
-else
-  for file in ${rundir}/*; do
-    echo "------------------------------"
-    echo "Content of '$file'"
-    echo "******************************"
-    cat $file
-    echo "******************************"
-  done
-  rm -rf ${rundir} || true
-fi
@@ -1,42 +0,0 @@
-#!/bin/bash
-
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-echo "Can only run if all the ${MACHINE_NAME} jobs passed"
-
-cd ${AUTOTEST_ROOT}/autotest || \
-  { echo "Invalid 'autotest' dir: ${AUTOTEST_ROOT}/autotest"; exit 1; }
-mkdir -p ${MACHINE_NAME}
-
-rundir="${MACHINE_NAME}/$(date +%Y-%m-%d)-gitlab-ci-${CI_COMMIT_REF_SLUG}"
-rundir=$(${CI_PROJECT_DIR}/.gitlab/scripts/safe_create_rundir $rundir)
-
-printf "%s\n" "The 'build-and-test' jobs on ${MACHINE_NAME} were SUCCESSFUL." \
-  "Pipeline URL:" "$CI_PIPELINE_URL" > ${rundir}/gitlab.out
-
-msg="GitLab CI log for build-and-test on ${MACHINE_NAME} ($(date +%Y-%m-%d))"
-
-if [[ "$AUTOTEST_COMMIT" != "NO" ]]; then
-  git pull && \
-  git add ${rundir} && \
-  git commit -m "${msg}" && \
-  git push origin master
-else
-  for file in ${rundir}/*; do
-    echo "------------------------------"
-    echo "Content of '$file'"
-    echo "******************************"
-    cat $file
-    echo "******************************"
-  done
-  rm -rf ${rundir} || true
-fi
@@ -1,42 +0,0 @@
-#!/bin/bash
-
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# This script takes a seed for a directory name and appends it with a counter
-# incremented until it can create a new directory with it.
-
-# Usage:
-#
-# Expects 1 argument: a string that is use as a seed for the directory name.
-#
-# > rundir="desired_name"
-# > rundir=$(./safe_create_rundir $rundir)
-
-set -o errexit
-set -o nounset
-
-rundir=${1:-""}
-if [[ -z ${rundir} ]]; then
-  >&2 echo "The script expects a string as argument for directory creation."
-  exit 1
-fi
-
-if ! mkdir ${rundir}; then
-  n=1
-  while ! mkdir ${rundir}_${n}
-  do
-    n=$((n+1))
-  done
-  rundir=${rundir}_${n}
-fi
-
-echo $rundir
@@ -10,53 +10,8 @@

 Version 4.3.1 (development)
 ===========================
-
- Add hipSPARSE support for sparse mat-vec multiplications.
-
- Added support for using the HYPRE library built with HIP support. Similar to
-  the HYPRE + CUDA support added earlier, most of the MFEM examples and miniapps
-  work transparently with HYPRE + HIP builds. This includes the BoomerAMG, AMS,
-  and ADS solvers.
-
- More explicit and consistent formating of the output of iterative solvers
-  with the new IterativeSolver::PrintLevel options. See linalg/solvers.hpp.
-
- Added a miniapp for PDE-based extrapolation of finite element functions. See
-  miniapps/shifted/extrapolate.cpp.
-
- Added support for automatic differentiation. Users can select between native
-  implementation and external library implementation during configuration. One
-  parallel and two serial examples are implemented in the miniapps/autodiff/
-  directory.
-
- GridFunctionCoefficient (and the related vector, gradient, divergence, and
-  curl classes) now work properly with LORDiscretization and LORSolver.
-
- Added support for mesh preprocessing to resolve fine scale problem data
-  before simulation. This feature uses adaptive mesh refinement to control the
-  associated data oscillation error. See the new Example 30/30p.
-
- Switched from Artistic Style (astyle) version 2.05.1 to version 3.1 for code
-  formatting. See the "make style" target.
-
- Split the fem/fe.?pp files into separate files in the new fem/fe/ directory
-  to simplify and clarify the organization of FiniteElement classes.
-
 - Added support for hr-adaptivity using TMOP-based error estimator.

- Coefficient::SetTime now propagates the new time into internally stored
-  Coefficient objects.
-
- Added initial support for google-benchmarks in the tests/benchmarks directory.
-  It can be enabled with MFEM_USE_BENCHMARK=YES.
-
- Added Binder (mybinder.org) configuration files for C++ MFEM Jupyter Notebooks
-  with inline GLVis visualization as well as a new examples/jupyter/ directory
-  with a sample notebook based on Example 1. Implementation based on xeus-cling,
-  github.com/jupyter-xeus/xeus-cling + xeus-glvis, github.com/GLVis/xeus-glvis.
-
- Added 'double' atomicAdd implementation for previous versions of CUDA.
-
 - Adding lowest order Nedelec and Raviart-Thomas basis functions on wedge
  shaped elements.

@@ -72,35 +27,6 @@ Version 4.3.1 (development)
  functions on wedges and pyramids which are not amenable to reordering. The
  ReorientTetMesh method of the Mesh and ParMesh classes has been deprecated.

- Gmsh meshes where all elements have zero physical tag (the default Gmsh
-  output format if no physical groups are defined) are now successfully loaded,
-  and elements are reassigned attribute number 1.
-
- Added new miniapps that use the ParELAG library, its hybrid smoothers, and the
-  hierarchy of spaces created by the element-based AMG (AMGe) methodology in
-  ParELAG to build multigrid solvers for H(curl) and H(div) forms. See the
-  miniapps/parelag directory for more details.
-
- Fixed several MinGW build issues on Windows.
-
- Remove the 'u' flag in the ar command, to update all files in the archive,
-  avoiding file name collisions from different subdirectories.
-
- Added initial TMOP-based capabilities for surface fitting and tangential
-  relaxation in the mesh-optimizer and pmesh-optimizer miniapps.
-
- Added ParMesh Adjaceny Set (adjset) creation support to the Conduit Mesh
-  Blueprint MFEM wrapper functions in ConduitDataCollection.
-
- `HypreParVector` and `Vector` now support move semantics, and the copy
-  constructor for `HypreParVector` now copies the local vector data.
-
- The HPC versions of ex1 and ex1p (in miniapps/performance) now support
-  runtime selection of either 2D or 3D meshes.
-
- Added ParaView visualization of `QuadratureFunction` fields, through both
-  `QuadratureFunction::SaveVTU` and `ParaViewDataCollection::RegisterQField`.
-

 Version 4.3, released on July 29, 2021
 ======================================
@@ -16,6 +16,9 @@ set(USER_CONFIG "${CMAKE_CURRENT_SOURCE_DIR}/config/user.cmake" CACHE PATH

 # Require C++11 and disable compiler-specific extensions
 set(CMAKE_CXX_STANDARD 11)
+if (MFEM_USE_GINKGO)
+  set(CMAKE_CXX_STANDARD 14)
+endif()
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
 set(CMAKE_CXX_EXTENSIONS OFF)

@@ -81,20 +84,12 @@ if (MFEM_USE_STRUMPACK)
  # Just needed to find the MPI_Fortran libraries to link with
  set(XSDK_ENABLE_Fortran ON)
 endif()
-if (MFEM_USE_GINKGO AND ("${CMAKE_CXX_STANDARD}" LESS "14"))
-  set(CMAKE_CXX_STANDARD 14)
-endif()

 # Include xSDK default CMake file.
 include("${CMAKE_CURRENT_SOURCE_DIR}/config/XSDKDefaults.cmake")

 # Enable languages.
 enable_language(CXX)
-if (MINGW)
-   # MinGW GCC does not expose the functions jn/_jn, yn/_yn (used in Example
-   # 25/25p) unless we use '-std=gnu++11':
-   set(CMAKE_CXX_EXTENSIONS ON)
-endif()
 if (MFEM_USE_CUDA)
   if (MFEM_USE_HIP)
      message(FATAL_ERROR " *** MFEM_USE_HIP cannot be combined with MFEM_USE_CUDA.")
@@ -180,17 +175,8 @@ else()
  set(MFEM_DEBUG OFF)
 endif()

-# AMD HIP
-if (MFEM_USE_HIP)
-  if (HIP_ARCH)
-    message(STATUS "Using HIP architecture: ${HIP_ARCH}")
-    set(GPU_TARGETS "${HIP_ARCH}" CACHE STRING "HIP targets to compile for")
-  endif()
-  if (ROCM_PATH)
-    list(INSERT CMAKE_PREFIX_PATH 0 ${ROCM_PATH})
-  endif()
-  find_package(HIP REQUIRED)
-  find_package(HIPSPARSE REQUIRED)
+if (WIN32)
+	add_definitions(-D_USE_MATH_DEFINES)
 endif()

 # MPI -> hypre; PETSc (optional)
@@ -265,11 +251,6 @@ if (MFEM_USE_OPENMP OR MFEM_USE_LEGACY_OPENMP)
  endif()
  find_package(OpenMP REQUIRED)
  set(OPENMP_LIBRARIES ${OpenMP_CXX_LIBRARIES})
-  if(APPLE)
-    # On macOS, the compiler needs additional help to find the <omp.h> header.
-    # See issue #2642 for more information.
-    include_directories(${OpenMP_CXX_INCLUDE_DIRS})
-  endif(APPLE)
 endif()

 # SuiteSparse (before SUNDIALS which may depend on KLU)
@@ -351,11 +332,11 @@ if (MFEM_USE_AMGX)
 endif()

 if (MFEM_USE_CONDUIT)
-    find_package(Conduit REQUIRED conduit relay blueprint)
+    find_package(Conduit REQUIRED conduit relay blueprint )
 endif()

 if (MFEM_USE_FMS)
-    find_package(FMS REQUIRED fms)
+    find_package(FMS REQUIRED fms )
 endif()

 # Axom/Sidre
@@ -367,7 +348,7 @@ endif()
 if (MFEM_USE_PUMI)
  # If PUMI_DIR was specified, only link to that directory,
  # i.e. don't link to another installation in /usr/lib by mistake
-  find_package(SCOREC 2.2.6 REQUIRED OPTIONAL_COMPONENTS gmi_sim
+  find_package(SCOREC 2.1.0 REQUIRED OPTIONAL_COMPONENTS gmi_sim
    CONFIG PATHS ${PUMI_DIR} NO_DEFAULT_PATH)
  if (SCOREC_FOUND)
    # Define a header file with the MFEM_USE_SIMMETRIX preprocessor variable
@@ -385,12 +366,6 @@ if (MFEM_USE_HIOP)
  # find_package updates HIOP_FOUND, HIOP_INCLUDE_DIRS, HIOP_LIBRARIES
 endif()

-# CoDiPack package
-if (MFEM_USE_CODIPACK)
-  find_package(CODIPACK REQUIRED)
-  # find_package updates CODIPACK_FOUND, CODIPACK_INCLUDE_DIRS, CODIPACK_LIBRARIES
-endif()
-
 # OCCA
 if (MFEM_USE_OCCA)
   find_package(OCCA REQUIRED)
@@ -406,16 +381,23 @@ if (MFEM_USE_UMPIRE)
   find_package(UMPIRE REQUIRED)
 endif()

-# GOOGLE-BENCHMARK
-if (MFEM_USE_BENCHMARK)
-   find_package(Benchmark REQUIRED)
-endif()
-
 # Caliper
 if (MFEM_USE_CALIPER)
  find_package(Caliper REQUIRED)
 endif()

+# AMD HIP
+if (MFEM_USE_HIP)
+  find_package(HIP REQUIRED)
+  if (HIP_ARCH)
+    message(STATUS "Using HIP architecture: ${HIP_ARCH}")
+    list(APPEND HIP_HIPCC_FLAGS "--amdgpu-target=${HIP_ARCH}")
+    if (MFEM_USE_GINKGO)
+      list(APPEND HIP_HIPCC_FLAGS "-std=c++14")
+    endif()
+  endif()
+endif()
+
 # ADIOS2 for parallel I/O
 if (MFEM_USE_ADIOS2)
   find_package(ADIOS2 REQUIRED)
@@ -427,11 +409,6 @@ if (MFEM_USE_MKL_CPARDISO)
  endif()
 endif()

-# PARELAG
-if (MFEM_USE_PARELAG)
-   find_package(PARELAG REQUIRED)
-endif()
-
 # MFEM_TIMER_TYPE
 if (NOT DEFINED MFEM_TIMER_TYPE)
  if (APPLE)
@@ -455,11 +432,10 @@ endif()
 #    With newer versions of SuiteSparse which include METIS header using 64-bit
 #    integers, the METIS header (with 32-bit indices, as used by mfem) needs to
 #    be before SuiteSparse.
-set(MFEM_TPLS OPENMP HYPRE BLAS LAPACK SuperLUDist METIS SuiteSparse SUNDIALS
-    PETSC SLEPC MESQUITE MUMPS STRUMPACK AXOM FMS CONDUIT Ginkgo GNUTLS GSLIB
-    NETCDF MPFR PUMI HIOP POSIXCLOCKS MFEMBacktrace ZLIB OCCA CEED RAJA UMPIRE
-    ADIOS2 CUSPARSE MKL_CPARDISO AMGX CALIPER CODIPACK BENCHMARK PARELAG
-    MPI_CXX HIP HIPSPARSE)
+set(MFEM_TPLS MPI_CXX OPENMP HYPRE BLAS LAPACK SuperLUDist METIS SuiteSparse SUNDIALS PETSC
+    SLEPC MESQUITE MUMPS STRUMPACK AXOM FMS CONDUIT Ginkgo GNUTLS GSLIB NETCDF
+    MPFR PUMI HIOP POSIXCLOCKS MFEMBacktrace ZLIB OCCA CEED RAJA UMPIRE ADIOS2
+    CUSPARSE MKL_CPARDISO AMGX CALIPER)

 # Add all *_FOUND libraries in the variable TPL_LIBRARIES.
 set(TPL_LIBRARIES "")
@@ -502,6 +478,8 @@ endforeach()

 if (MFEM_USE_CUDA)
  set_source_files_properties(${SOURCES} PROPERTIES LANGUAGE CUDA)
+elseif(MFEM_USE_HIP)
+  set_source_files_properties(${SOURCES} PROPERTIES HIP_SOURCE_PROPERTY_FORMAT TRUE)
 endif()

 add_subdirectory(config)
@@ -522,9 +500,13 @@ set(MFEM_INSTALL_DIR ${CMAKE_INSTALL_PREFIX} CACHE PATH
 # Declaring the library
 mfem_add_library(mfem ${SOURCES} ${HEADERS} ${MASTER_HEADERS})
 # message(STATUS "TPL_LIBRARIES = ${TPL_LIBRARIES}")
-target_link_libraries(mfem PUBLIC ${TPL_LIBRARIES})
+if (CMAKE_VERSION VERSION_GREATER 2.8.11)
+  target_link_libraries(mfem PUBLIC ${TPL_LIBRARIES})
+else()
+  target_link_libraries(mfem ${TPL_LIBRARIES})
+endif()
 if (MINGW)
-  target_link_libraries(mfem PRIVATE ws2_32)
+  target_link_libraries(mfem ws2_32)
 endif()
 set_target_properties(mfem PROPERTIES VERSION "${mfem_VERSION}")
 set_target_properties(mfem PROPERTIES SOVERSION "${mfem_VERSION}")
@@ -563,21 +545,15 @@ endif()
 set(MFEM_CUSTOM_TARGET_PREFIX CACHE STRING "")

 #-------------------------------------------------------------------------------
-# Examples, miniapps, benchmarks and testing
+# Examples, miniapps, and testing
 #-------------------------------------------------------------------------------

-# Enable testing and benchmarks if required
+# Enable testing if required
 if (MFEM_ENABLE_TESTING)
  enable_testing()
  set(MFEM_ALL_TESTS_TARGET_NAME tests)
  add_mfem_target(${MFEM_ALL_TESTS_TARGET_NAME} OFF)
  add_subdirectory(tests EXCLUDE_FROM_ALL)
-
-  # Create a target for all benchmarks and, optionally, enable it.
-  set(MFEM_ALL_BENCHMARKS_TARGET_NAME benchmarks)
-  add_mfem_target(${MFEM_ALL_BENCHMARKS_TARGET_NAME}
-                  ${MFEM_ENABLE_GOOGLE_BENCHMARKS})
-  add_subdirectory(tests/benchmarks EXCLUDE_FROM_ALL)
 endif()

 # Define a target that all examples and miniapps will depend on.
@@ -587,11 +563,7 @@ add_custom_target(${MFEM_EXEC_PREREQUISITES_TARGET_NAME})
 # Create a target for all examples and, optionally, enable it.
 set(MFEM_ALL_EXAMPLES_TARGET_NAME examples)
 add_mfem_target(${MFEM_ALL_EXAMPLES_TARGET_NAME} ${MFEM_ENABLE_EXAMPLES})
-if (MFEM_ENABLE_EXAMPLES)
-  add_subdirectory(examples) #install examples if enabled
-else()
-  add_subdirectory(examples EXCLUDE_FROM_ALL)
-endif()
+add_subdirectory(examples EXCLUDE_FROM_ALL)

 # Create a target for all miniapps and, optionally, enable it.
 set(MFEM_ALL_MINIAPPS_TARGET_NAME miniapps)
@@ -601,7 +573,6 @@ add_subdirectory(miniapps EXCLUDE_FROM_ALL)
 # Target to build all executables, i.e. everything.
 add_custom_target(exec)
 add_dependencies(exec
-  ${MFEM_ALL_BENCHMARKS_TARGET_NAME}
  ${MFEM_ALL_EXAMPLES_TARGET_NAME}
  ${MFEM_ALL_MINIAPPS_TARGET_NAME}
  ${MFEM_ALL_TESTS_TARGET_NAME})
@@ -1,133 +0,0 @@
-# MFEM Code of Conduct
-
-## Our Pledge
-
-We as members, contributors, and leaders pledge to make participation in our
-community a harassment-free experience for everyone, regardless of age, body
-size, visible or invisible disability, ethnicity, sex characteristics, gender
-identity and expression, level of experience, education, socio-economic status,
-nationality, personal appearance, race, caste, color, religion, or sexual
-identity and orientation.
-
-We pledge to act and interact in ways that contribute to an open, welcoming,
-diverse, inclusive, and healthy community.
-
-## Our Standards
-
-Examples of behavior that contributes to a positive environment for our
-community include:
-
-* Demonstrating empathy and kindness toward other people
-* Being respectful of differing opinions, viewpoints, and experiences
-* Giving and gracefully accepting constructive feedback
-* Accepting responsibility and apologizing to those affected by our mistakes,
-  and learning from the experience
-* Focusing on what is best not just for us as individuals, but for the overall
-  community
-
-Examples of unacceptable behavior include:
-
-* The use of sexualized language or imagery, and sexual attention or advances of
-  any kind
-* Trolling, insulting or derogatory comments, and personal or political attacks
-* Public or private harassment
-* Publishing others' private information, such as a physical or email address,
-  without their explicit permission
-* Other conduct which could reasonably be considered inappropriate in a
-  professional setting
-
-## Enforcement Responsibilities
-
-Community leaders are responsible for clarifying and enforcing our standards of
-acceptable behavior and will take appropriate and fair corrective action in
-response to any behavior that they deem inappropriate, threatening, offensive,
-or harmful.
-
-Community leaders have the right and responsibility to remove, edit, or reject
-comments, commits, code, wiki edits, issues, and other contributions that are
-not aligned to this Code of Conduct, and will communicate reasons for moderation
-decisions when appropriate.
-
-## Scope
-
-This Code of Conduct applies within all community spaces, and also applies when
-an individual is officially representing the community in public spaces.
-Examples of representing our community include using an official e-mail address,
-posting via an official social media account, or acting as an appointed
-representative at an online or offline event.
-
-## Enforcement
-
-Instances of abusive, harassing, or otherwise unacceptable behavior may be
-reported to the community leaders responsible for enforcement at mfem@llnl.gov.
-All complaints will be reviewed and investigated promptly and fairly.
-
-All community leaders are obligated to respect the privacy and security of the
-reporter of any incident. Anyone involved in the reported behavior will recuse
-themselves from the investigation and decision making about the resolution of
-the complaint.
-
-## Enforcement Guidelines
-
-Community leaders will follow these Community Impact Guidelines in determining
-the consequences for any action they deem in violation of this Code of Conduct:
-
-### 1. Correction
-
-**Community Impact**: Use of inappropriate language or other behavior deemed
-unprofessional or unwelcome in the community.
-
-**Consequence**: A private, written warning from community leaders, providing
-clarity around the nature of the violation and an explanation of why the
-behavior was inappropriate. A public apology may be requested.
-
-### 2. Warning
-
-**Community Impact**: A violation through a single incident or series of
-actions.
-
-**Consequence**: A warning with consequences for continued behavior. No
-interaction with the people involved, including unsolicited interaction with
-those enforcing the Code of Conduct, for a specified period of time. This
-includes avoiding interactions in community spaces as well as external channels
-like social media. Violating these terms may lead to a temporary or permanent
-ban.
-
-### 3. Temporary Ban
-
-**Community Impact**: A serious violation of community standards, including
-sustained inappropriate behavior.
-
-**Consequence**: A temporary ban from any sort of interaction or public
-communication with the community for a specified period of time. No public or
-private interaction with the people involved, including unsolicited interaction
-with those enforcing the Code of Conduct, is allowed during this period.
-Violating these terms may lead to a permanent ban.
-
-### 4. Permanent Ban
-
-**Community Impact**: Demonstrating a pattern of violation of community
-standards, including sustained inappropriate behavior, harassment of an
-individual, or aggression toward or disparagement of classes of individuals.
-
-**Consequence**: A permanent ban from any sort of public interaction within the
-community.
-
-## Attribution
-
-This Code of Conduct is adapted from the [Contributor Covenant][homepage],
-version 2.1, available at
-[https://www.contributor-covenant.org/version/2/1/code_of_conduct.html][v2.1].
-
-Community Impact Guidelines were inspired by
-[Mozilla's code of conduct enforcement ladder][Mozilla CoC].
-
-For answers to common questions about this code of conduct, see the FAQ at
-[https://www.contributor-covenant.org/faq][FAQ]. Translations are available at
-[https://www.contributor-covenant.org/translations][translations].
-
-[homepage]: https://www.contributor-covenant.org
-[v2.1]: https://www.contributor-covenant.org/version/2/1/code_of_conduct.html
-[Mozilla CoC]: https://github.com/mozilla/diversity
-[FAQ]: https://www.contributor-covenant.org/faq
-[translations]: https://www.contributor-covenant.org/translations
@@ -21,9 +21,6 @@ documentation; new examples and miniapps; HPC performance improvements; etc.
 MFEM is distributed under the terms of the BSD-3 license. All new contributions
 must be made under this license.

-Note also that MFEM has a [Code of Conduct](CODE_OF_CONDUCT.md). By participating
-in the MFEM community, you agree to abide by its rules.
-
 If you plan on contributing to MFEM, consider reviewing the
 [issue tracker](https://github.com/mfem/mfem/issues) first to check if a thread
 already exists for your desired feature or the bug you ran into. Use a pull
@@ -45,7 +42,6 @@ back to them before issuing pull requests:
  - [New Feature Development](#new-feature-development)
  - [Developer Guidelines](#developer-guidelines)
  - [Pull Requests](#pull-requests)
-  - [MFEM PR Rules](#mfem-pr-rules)
  - [Pull Request Checklist](#pull-request-checklist)
  - [Master/Next Workflow](#masternext-workflow)
  - [Releases](#releases)
@@ -71,9 +67,8 @@ Origin](#developers-certificate-of-origin-11) at the end of this file.*
  with regards to documentation and code styling.
 - Please do not commit large/binary files to the central repository (use a fork
  instead).
- Pull requests should be issued toward `mfem:master`. Make sure
-  to check the items off the [Pull Request Checklist](#pull-request-checklist) and
-  follow the [MFEM PR Rules](#mfem-pr-rules).
+- Pull requests  should be issued toward `mfem:master`. Make sure
+  to check the items off the [Pull Request Checklist](#pull-request-checklist).
 - When your contribution is fully working and ready to be reviewed, add
  the `ready-for-review` label.
 - PRs are treated similarly to journal submission with an "editor" assigning two
@@ -110,14 +105,12 @@ The MFEM source code has the following structure:
  │   ├── caliper
  │   ├── ginkgo
  │   ├── hiop
-  │   ├── jupyter
  │   ├── petsc
  │   ├── pumi
  │   ├── sundials
  |   └── superlu
  ├── fem
  │   ├── ceed
-  │   ├── fe
  │   ├── qinterp
  │   └── tmop
  ├── general
@@ -126,7 +119,6 @@ The MFEM source code has the following structure:
  ├── mesh
  ├── miniapps
  │   ├── adjoint
-  │   ├── autodiff
  │   ├── common
  │   ├── electromagnetics
  │   ├── gslib
@@ -134,7 +126,6 @@ The MFEM source code has the following structure:
  │   ├── mtop
  │   ├── navier
  │   ├── nurbs
-  │   ├── parelag
  │   ├── performance
  │   ├── shifted
  │   ├── solvers
@@ -332,22 +323,15 @@ Before you can start, you need a GitHub account, here are a few suggestions:
    change the code by default.

 - Code specifics
-  - All new public, protected, and private classes, methods, data members, and
-    functions have Doxygen-style documentation in source comments.
-  - In addition to arguments and functionality, documentation should include the
-    current limitations of the code, any background information that is
-    implicitly assumed in the implementation, and the ownership and lifetime
-    of data.
+  - All significant new classes, methods and functions have Doxygen-style
+    documentation in source comments.
  - Consistent code styling is enforced with `make style` in the top-level
    directory. This requires [Artistic Style](http://astyle.sourceforge.net) (we
-    specifically use version 3.1). See also the file `config/mfem.astylerc`.
+    specifically use version 2.05.1). See also the file `config/mfem.astylerc`.
  - Use `mfem::out` and `mfem::err` instead of `std::cout` and `std::cerr` in
    internal library code. (You can use `std` in examples and miniapps.)
  - When manually resolving conflicts during a merge, make sure to mention the
    conflicted files in the commit message.
-  - All significant new features and changes should be documented in CHANGELOG.
-  - New examples and miniapps should have documentation on the MFEM webpage.
-

 ### Pull Requests

@@ -413,83 +397,6 @@ Before you can start, you need a GitHub account, here are a few suggestions:
 - If triggered, track the status of the LLNL GitLab tests. If failing, ask
  one of the _LLNL developers_ for details.

-
-### MFEM PR Rules
-
-The Pull Request (PR) approval process in MFEM is similar to the approval of papers in a peer-reviewed journal. In particular:
-
-1. There is an MFEM board of "editors" that evaluates new PRs and assigns "reviewers" for each PR.
-
-2. The assigned reviewers are responsible to carefully review and test the proposed PR.
-
-3. A PR can be (manually) merged in the *next* branch only if 2 of the assigned reviewers have approved it and it has passed internal testing. This merge can be performed by any of the assigned reviewers or by any of the editors.
-
-4. A PR can be merged in the *master* branch only if it has been tested successfully for a week in *next* and an editor has (optionally) taken a final look. This merge can be performed only by one of the editors.
-
-#### Responsibilities of Editors
-
-The current list of MFEM editors is:
-
- @v-dobrev (Veselin Dobrev)
- @tzanio (Tzanio Kolev)
- @pazner (Will Pazner)
- @mlstowell (Mark Stowell)
-
-**The responsibilities of the editors are:**
-
-1. To assign appropriate milestone and labels for new PRs, e.g. *bugfix*, *minor*, *api-change*, *high-impact*, etc.
-
-2. To assign at least 2 reviewers for new PRs. An editor can also be a reviewer. The editor, reviewers, and author should be listed as "Assignees" on the GitHub PR page. After assignment, the `in-review` label should be added.
-
-3. To complete the initial PR evaluation and assignments in a timely manner: 1 week from submission.
-
-4. To assist reviewers when they need help with their reviews (but also to stay out of the way when they don't).
-
-5. To remind the reviewers about timely completion of their review.
-
-6. To take a final look and complete the PR merge in *master*. The final look step is optional and shouldn't take more than 3 days.
-
-7. The assignment of bugfixes should be expedited proportional to their importance, e.g. in some cases the editor can assign much shorter review window.
-
-#### Responsibilities of Reviewers
-
-Everyone on the MFEM team can be asked to serve as a reviewer on a PR in their area of expertise.
-
-**The responsibilities of the reviewers are:**
-
-1. To let the editors know if the proposed assignment is not a good match for them.
-
-2. To communicate with the PR author, provide feedback and work with them to resolve issues.
-
-3. To ensure the quality of the PR by making sure that the code adheres to the [Developer Guidelines](#developer-guidelines), e.g. all methods, data members, and functions have documentation, including data ownership and lifetime, new examples/miniapps have a corresponding PR in mfem/web, major features have `CHANGELOG` entries, etc.
-
-3. To seek help from the editors in case of difficulties.
-
-4. To complete the review in a timely manner: 3 weeks from assignment.
-
-5. To test the PR thoroughly before merging in *next*. The PR author is also encouraged to perform testing and inform the reviewers about the results.
-
-6. To monitor the PR impact on the testing in the *next* branch and alert the editors that the PR is ready for merging in *master*.
-
-7. The review of bugfixes should be expedited proportional to their importance. The review window can be much less than three weeks in such cases.
-
-#### Responsibilities of Authors
-
-Authors should clearly indicate when a PR is ready for review (before that the PR should be marked as `Draft` or `[WIP]`).
-
-**The responsibilities of the authors are:**
-
-1. To follow the instructions and PR checklist in the `CONTRIBUTING.md` document in the MFEM repository.
-
-2. To respond to reviewer feedback in a timely manner.
-
-3. Authors are encouraged to perform testing and inform the reviewers about the results.
-
-4. Authors can use the "Reviewers" section of the GitHub PR page to suggest reviewers, but the "Assignees" section will show who the editor has assigned to do the reviews.
-
-5. To indicate when the PR is ready for review by adding the `ready-for-review` label.
-
-
 ### Pull Request Checklist

 Before a PR can be merged, it should satisfy the following:
@@ -543,9 +450,7 @@ Before a PR can be merged, it should satisfy the following:
     - [ ] The miniapps go at the end of the page, and are usually listed only under a specific "Application (PDE)" category.
     - [ ] Add a short description of the miniapp in the "Extensive Examples" section of `features.md`.
 - [ ] New capability:
-   - [ ] All new public, protected, and private classes, methods, data members, and functions have full Doxygen-style documentation in source comments. Documentation should include descriptions of member data, function arguments and return values, template parameters, and prerequisites for calling new functions.
-   - [ ] Pointer arguments and return values must specify whether ownership is being transferred or lent with the call.
-   - [ ] Any new functions should include descriptions of their intended use e.g. for internal use only, user-facing, etc., along with references to example code whenever possible/appropriate.
+   - [ ] All significant new classes, methods and functions have Doxygen-style documentation in source comments.
   - [ ] Consider adding new sample runs in existing examples to highlight the new capability.
   - [ ] Consider saving cool simulation pictures with the new capability in the Confluence gallery (LLNL only) or submitting them, via pull request, to the gallery section of the `mfem/web` repo.
   - [ ] If this is a major new feature, consider mentioning it in the short summary inside `README` *(rare)*.
@@ -556,7 +461,6 @@ Before a PR can be merged, it should satisfy the following:
 - [ ] (LLNL only) After merging:
   - [ ] Update internal tests to include the new features.

-
 ### Master/Next Workflow

 MFEM uses a `master`/`next`-branch workflow as described below:
@@ -648,10 +552,8 @@ MFEM uses a `master`/`next`-branch workflow as described below:
    - Update version and shortlinks in `src/index.md` and `src/download.md`.
    -  Use [cloc-1.62.pl](http://cloc.sourceforge.net/) and `ls -lh` to estimate the SLOC and the tarball size in `src/download.md`.

-
 ## LLNL Workflow

-
 ### Mirroring on Bitbucket

 - The GitHub `master` and `next` branches are mirrored to the LLNL institutional
@@ -671,17 +573,16 @@ MFEM uses a `master`/`next`-branch workflow as described below:
  - `mfem:gh-next` -- Bleeding-edge development version, may be broken, use at
     your own risk.

-
 ### Mirroring on GitLab

 - MFEM repository is also mirrored on the LLNL GitLab instance, in a
  semi-automated manner.

 - This instance is meant to complete CI testing with tests on Livermore
-  Computing systems. GitLab pipeline status is reported in the corresponding
+  Computing systems. Gitlab pipeline status is reported in the corresponding
  GitHub pull request.

- In GitLab pipelines, TPLs (dependencies) are built using Spack, driven by Uberenv.
+- In Gitlab pipelines, TPLs (dependencies) are built using Spack, driven by Uberenv.

 - No change to the MFEM repo can be made on this instance.

@@ -694,7 +595,6 @@ In addition, developers can set local git hooks to run some quick checks on
 commit or push, see the [README](config/githooks/README.md) in the `config/githooks`
 directory.

-
 ### Linux and Mac smoke tests
 We use GitHub Actions to drive the default tests on the `master` and `next`
 branches. See the `.github/workflows` files and the logs at
@@ -706,7 +606,6 @@ constraint on jobs. Two virtual machines are configured - Mac (OS X) and Linux.
 - Tests on the `master` branch are triggered whenever a PR is issued on this branch.
 - Tests on the `next` branch are currently scheduled to run each night.

-
 ### Windows smoke test
 We use Appveyor to test building with the MS Visual C++ compiler in a Windows
 environment, as well as to test the CMake build. See the `.appveyor` file and the
@@ -716,7 +615,6 @@ build logs at
 CMake is used to generate the MSVC Project files and drive the build.  A release
 and debug build is performed with a simple run of `ex1` to verify the executable.

-
 ### Tests at LLNL

 - We mirror the `master` and `next` branches internally (to `gh-master` and
@@ -37,7 +37,7 @@ as CUDA, HIP, OCCA, OpenMP and RAJA.
  https://developer.nvidia.com/cuda-toolkit

 - HIP support requires an AMD GPU and an installation of the ROCm software stack
-  https://rocmdocs.amd.com
+  https://rocm.github.io/ROCmInstall.html#installing-from-amd-rocm-repositories

 - OCCA support requires the OCCA library
  https://libocca.org
@@ -58,8 +58,7 @@ following package managers:

 - Spack, https://github.com/spack/spack
 - OpenHPC, http://openhpc.community
- Conda-forge, https://conda-forge.org (pre-built binaries linked with
-  OpenMPI/MPICH, hypre, and METIS)
+- Conda-forge, https://conda-forge.org (pre-built binaries linked with OpenMPI/MPICH, hypre, and METIS)
 - Homebrew/Science, https://github.com/Homebrew/homebrew-science (deprecated)

 We also recommend downloading and building the MFEM-based GLVis visualization
@@ -79,7 +78,7 @@ Parallel build:

 CUDA build:
   make cuda -j 4
-   (build for a specific compute capability: 'make cuda -j 4 CUDA_ARCH=sm_70')
+   (build for a specific compute capability: 'make cuda -j 4 CUDA_ARCH=sm_30')

 HIP build:
   make hip -j 4
@@ -460,22 +459,10 @@ MFEM_USE_UMPIRE = YES/NO
   discovery, provision, and management of memory on machines with multiple
   memory devices like NUMA and GPUs.

-MFEM_USE_BENCHMARK = YES/NO
-   Enables support for Google Benchmark, a library to support the benchmarking
-   of functions, in the tests/benchmarks directory.
-
 MFEM_USE_HIOP = YES/NO
   Enable the usage of HiOp (https://github.com/LLNL/hiop) in MFEM. HiOp is an
   HPC solver for nonlinear optimization problems.

-MFEM_USE_CODIPACK = YES/NO
-  Enable automatic differentiation using the CoDiPack library.
-  www.scicomp.uni-kl.de/codi/
-
-MFEM_USE_ADFORWARD = YES/NO
-  Enable forward mode for AD packages. This option is valid
-  only if the AD package supports two modes (backward/forward).
-
 MFEM_USE_CUDA = YES/NO
   Enables support for CUDA devices in MFEM. CUDA is a parallel computing
   platform and programming model for general computing on graphical processing
@@ -536,11 +523,6 @@ MFEM_USE_FMS = YES/NO
   convetion routines between FMS's FmsDataCollection structure and MFEM's
   DataCollection class, see the header file fem/fmsconvert.hpp.

-MFEM_USE_PARELAG = YES/NO
-   Enables the miniapps that use the ParELAG library. MFEM does not currently
-   use ParELAG. In fact, ParELAG is dependent on MFEM. Therefore, this option
-   currently only concerns the miniapps.
-
 MFEM_BUILD_TAG = (any value)
   An optional tag to characterize the build. Exported to config/config.mk.
   Can be used to identify the MFEM build from other makefiles.
@@ -567,7 +549,7 @@ The specific libraries and their options are:
  Options: HYPRE_OPT, HYPRE_LIB.
  Versions: HYPRE >= 2.10.0b  (HYPRE built without CUDA)
            HYPRE >= 2.20.0   (HYPRE built with '--enable-mixedint')
-            HYPRE >= 2.22.1   (HYPRE built with CUDA or HIP)
+            HYPRE >= 2.22.1   (HYPRE built with CUDA)

 - METIS, used when MFEM_USE_METIS = YES. If using METIS 5, set
  MFEM_USE_METIS_5 = YES (default is to use METIS 4).
@@ -640,8 +622,7 @@ The specific libraries and their options are:
  C++ compiler that supports the C++-14 standard. For additional requirements
  and dependencies of specific modules, see the Ginkgo webpage below.
  URL: https://ginkgo-project.github.io
-  Options: GINKGO_OPT, GINKGO_LIB, GINKGO_DIR, GINKGO_BUILD_TYPE (Release or
-           Debug).
+  Options: GINKGO_OPT, GINKGO_LIB, GINKGO_DIR, GINKGO_BUILD_TYPE (Release or Debug).
  Versions: Ginkgo >= 1.4.0.

 - AmgX (optional), used when MFEM_USE_AMGX = YES.
@@ -706,17 +687,12 @@ The specific libraries and their options are:
  URL: https://scorec.rpi.edu/pumi
       https://github.com/SCOREC/core
  Options: PUMI_OPT, PUMI_LIB.
-  Versions: PUMI >= 2.2.6.
+  Versions: PUMI == 2.2.3.

 - HiOp (optional), used when MFEM_USE_HIOP = YES.
  URL: https://github.com/LLNL/hiop
  Options: HIOP_OPT, HIOP_LIB.
-  Versions: HIOP >= 0.4.6.
-
- CoDiPack (optiobal), used with MFEM_USE_CODIPACK = YES
-  URL: https://www.scicomp.uni-kl.de/codi/
-  Options: CODIPACK_OPT
-  Versions: 1.9.3
+  Versions: HIOP >= 0.4.

 - GSLIB (optional), used when MFEM_USE_GSLIB = YES. The gslib library must be
  built prior to the MFEM build, as follows: download gslib-1.0.7, untar it at
@@ -739,7 +715,7 @@ The specific libraries and their options are:
  Versions: CUDA >= 10.1.168.

 - HIP (optional), used when MFEM_USE_HIP = YES.
-  URL: https://rocmdocs.amd.com
+  URL: https://rocm.github.io/ROCmInstall.html
  Options: HIP_CXX, HIP_ARCH, HIP_OPT, HIP_LIB.

 - OCCA (optional), used when MFEM_USE_OCCA = YES.
@@ -754,10 +730,10 @@ The specific libraries and their options are:
  Versions: libCEED >= 0.8.

 - RAJA (optional), used when MFEM_USE_RAJA = YES.
-  Beginning with MFEM v4.3, only RAJA v0.14.0+ is supported.
+  Beginning with MFEM v4.3, only RAJA v0.13.0+ is supported.
  URL: https://github.com/LLNL/RAJA
  Options: RAJA_DIR, RAJA_OPT, RAJA_LIB.
-  Versions: RAJA >= 0.14.0.
+  Versions: RAJA >= 0.13.0.

 - Caliper (optional), used when MFEM_USE_CALIPER = YES.
  URL: https://github.com/LLNL/Caliper
@@ -768,12 +744,7 @@ The specific libraries and their options are:
  Umpire requires camp when the Umpire version is >= 3.0.0.
  URL: https://github.com/LLNL/Umpire
  Options: UMPIRE_DIR, UMPIRE_OPT, UMPIRE_LIB.
-  Versions: Umpire >= 3.0.0.
-
- Benchmark, used when MFEM_USE_BENCHMARK = YES.
-  URL: https://github.com/google/benchmark
-  Options: BENCHMARK_DIR, BENCHMARK_LIB.
-  Versions: Benchmark >= 1.5.6.
+  Versions: Umpire >= 2.0.0.

 - MPFR (optional), used when MFEM_USE_MPFR = YES.
  URL: http://mpfr.org, it depends on the GMP library: https://gmplib.org
@@ -795,10 +766,6 @@ The specific libraries and their options are:
  Options: FMS_OPT, FMS_LIB.
  Versions: FMS >= 0.2.

- ParELAG, used when MFEM_USE_PARELAG = YES.
-  URL: https://github.com/LLNL/parelag
-  Options: PARELAG_DIR, PARELAG_OPT, PARELAG_LIB.
-
 Building with CMake
 ===================
 The MFEM build system consists of two steps: configuration and compilation.
@@ -923,10 +890,7 @@ MFEM_USE_MPFR
 MFEM_USE_ZLIB
 MFEM_USE_PUMI
 MFEM_USE_HIOP
-MFEM_USE_CODIPACK
-MFEM_USE_ADFORWARD
 MFEM_USE_CUDA
-MFEM_USE_HIP
 MFEM_USE_OCCA
 MFEM_USE_CEED
 MFEM_USE_RAJA
@@ -934,8 +898,6 @@ MFEM_USE_UMPIRE
 MFEM_USE_SIDRE
 MFEM_USE_CALIPER
 MFEM_USE_FMS
-MFEM_USE_BENCHMARK
-MFEM_USE_PARELAG

 The following options are CMake specific:

@@ -985,15 +947,12 @@ The CMake build system adds auto-detection for the following packages/libraries:
 - POSIXCLOCKS
 - PUMI
 - HIOP
- - CoDiPack
 - OCCA
 - RAJA
 - UMPIRE
 - AXOM - Used when MFEM_USE_SIDRE is enabled
 - CALIPER
 - FMS
- - BENCHMARK
- - ParELAG

 The following built-in CMake packages are also used:

@@ -12,9 +12,6 @@ to enable high-performance scalable finite element discretization research and
 application development on a wide variety of platforms, ranging from laptops to
 supercomputers.

-We welcome contributions and feedback from the community. Please see the file
-CONTRIBUTING.md for additional details about our development process.
-
 * For building instructions, see the file INSTALL, or type "make help".

 * Copyright and licensing information can be found in files LICENSE and NOTICE.
@@ -22,6 +19,9 @@ CONTRIBUTING.md for additional details about our development process.
 * The best starting point for new users interested in MFEM's features is to
  review the examples and miniapps at https://mfem.org/examples.

+* Developers interested in contributing to the library, should read the
+  instructions and documentation in the CONTRIBUTING.md file.
+
 Conceptually, MFEM can be viewed as a finite element toolbox that provides the
 building blocks for developing finite element algorithms in a manner similar to
 that of MATLAB for linear algebra methods. In particular, MFEM provides support
@@ -283,11 +283,3 @@ ENDIF()
 IF (DEFINED TPL_ENABLE_UMPIRE)
  SET(MFEM_USE_UMPIRE ${TPL_ENABLE_UMPIRE} CACHE BOOL "Enable Umpire" FORCE)
 ENDIF()
-
-IF (DEFINED TPL_ENABLE_BENCHMARK)
-  SET(MFEM_USE_BENCHMARK ${TPL_ENABLE_BENCHMARK} CACHE BOOL "Enable Google-Benchmark" FORCE)
-ENDIF()
-
-IF (DEFINED TPL_ENABLE_PARELAG)
-  SET(MFEM_USE_PARELAG ${TPL_ENABLE_PARELAG} CACHE BOOL "Enable ParELAG" FORCE)
-ENDIF()
@@ -54,11 +54,7 @@ set(MFEM_USE_CEED @MFEM_USE_CEED@)
 set(MFEM_USE_UMPIRE @MFEM_USE_UMPIRE@)
 set(MFEM_USE_SIMD @MFEM_USE_SIMD@)
 set(MFEM_USE_ADIOS2 @MFEM_USE_ADIOS2@)
-set(MFEM_USE_CODIPACK @MFEM_USE_CODIPACK@)
-set(MFEM_USE_ADFORWARD @MFEM_USE_ADFORWARD@)
 set(MFEM_USE_CALIPER @MFEM_USE_CALIPER@)
-set(MFEM_USE_BENCHMARK @MFEM_USE_BENCHMARK@)
-set(MFEM_USE_PARELAG @MFEM_USE_PARELAG@)

 set(MFEM_CXX_COMPILER "@CMAKE_CXX_COMPILER@")
 set(MFEM_CXX_FLAGS "@CMAKE_CXX_FLAGS@")
@@ -175,13 +175,4 @@
 // Enable interface to the MKL CPardiso library.
 #cmakedefine MFEM_USE_MKL_CPARDISO

-// Use forward mode for automatic differentiation
-#cmakedefine MFEM_USE_ADFORWARD
-
-// Enable the use of the CoDiPack library for AD
-#cmakedefine MFEM_USE_CODIPACK
-
-// Enable MFEM functionality based on the Google Benchmark library.
-#cmakedefine MFEM_USE_BENCHMARK
-
 #endif // MFEM_CONFIG_HEADER
@@ -1,22 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# Defines the following variables:
-#   - BENCHMARK_FOUND
-#   - BENCHMARK_LIBRARIES
-#   - BENCHMARK_INCLUDE_DIRS
-
-include(MfemCmakeUtilities)
-mfem_find_package(Benchmark BENCHMARK BENCHMARK_DIR
-        "include" "benchmark/benchmark.h"
-        "lib" "benchmark"
-        "Paths to headers required by Google Benchmark."
-        "Libraries required by Google Benchmark.")
@@ -1,24 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# Automatic differentiation using the CoDiPack library.
-#  www.scicomp.uni-kl.de/codi/
-# Sets the following variables:
-#   - CODIPACK_FOUND
-#   - CODIPACK_INCLUDE_DIRS
-#   - CODIPACK_LIBRARIES
-
-include(MfemCmakeUtilities)
-mfem_find_package(CODIPACK CODIPACK CODIPACK_DIR
-  "include" "codi.h"
-  "lib" ""
-  "Paths to headers required by CODIPACK."
-  "Libraries required by CODIPACK.")
@@ -0,0 +1,692 @@
+###############################################################################
+# FindHIP.cmake
+###############################################################################
+include(CheckCXXCompilerFlag)
+###############################################################################
+# SET: Variable defaults
+###############################################################################
+# User defined flags
+set(HIP_HIPCC_FLAGS "" CACHE STRING "Semicolon delimited flags for HIPCC")
+set(HIP_HCC_FLAGS "" CACHE STRING "Semicolon delimited flags for HCC")
+set(HIP_CLANG_FLAGS "" CACHE STRING "Semicolon delimited flags for CLANG")
+set(HIP_NVCC_FLAGS "" CACHE STRING "Semicolon delimted flags for NVCC")
+mark_as_advanced(HIP_HIPCC_FLAGS HIP_HCC_FLAGS HIP_CLANG_FLAGS HIP_NVCC_FLAGS)
+
+set(_hip_configuration_types ${CMAKE_CONFIGURATION_TYPES} ${CMAKE_BUILD_TYPE} Debug MinSizeRel Release RelWithDebInfo)
+list(REMOVE_DUPLICATES _hip_configuration_types)
+foreach(config ${_hip_configuration_types})
+    string(TOUPPER ${config} config_upper)
+    set(HIP_HIPCC_FLAGS_${config_upper} "" CACHE STRING "Semicolon delimited flags for HIPCC")
+    set(HIP_HCC_FLAGS_${config_upper} "" CACHE STRING "Semicolon delimited flags for HCC")
+    set(HIP_CLANG_FLAGS_${config_upper} "" CACHE STRING "Semicolon delimited flags for CLANG")
+    set(HIP_NVCC_FLAGS_${config_upper} "" CACHE STRING "Semicolon delimited flags for NVCC")
+    mark_as_advanced(HIP_HIPCC_FLAGS_${config_upper} HIP_HCC_FLAGS_${config_upper} HIP_CLANG_FLAGS_${config_upper} HIP_NVCC_FLAGS_${config_upper})
+endforeach()
+option(HIP_HOST_COMPILATION_CPP "Host code compilation mode" ON)
+option(HIP_VERBOSE_BUILD "Print out the commands run while compiling the HIP source file.  With the Makefile generator this defaults to VERBOSE variable specified on the command line, but can be forced on with this option." OFF)
+mark_as_advanced(HIP_HOST_COMPILATION_CPP)
+
+###############################################################################
+# FIND: HIP and associated helper binaries
+###############################################################################
+
+get_filename_component(_IMPORT_PREFIX "${CMAKE_CURRENT_LIST_DIR}/../" REALPATH)
+
+# HIP is supported on Linux only
+if(UNIX AND NOT APPLE AND NOT CYGWIN)
+    # Search for HIP installation
+    if(NOT HIP_ROOT_DIR)
+        # Search in user specified path first
+        find_path(
+            HIP_ROOT_DIR
+            NAMES bin/hipconfig
+            PATHS
+            "$ENV{ROCM_PATH}/hip"
+            ENV HIP_PATH
+            ${_IMPORT_PREFIX}
+            /opt/rocm/hip
+            DOC "HIP installed location"
+            NO_DEFAULT_PATH
+            )
+        if(NOT EXISTS ${HIP_ROOT_DIR})
+            if(HIP_FIND_REQUIRED)
+                message(FATAL_ERROR "Specify HIP_ROOT_DIR")
+            elseif(NOT HIP_FIND_QUIETLY)
+                message("HIP_ROOT_DIR not found or specified")
+            endif()
+        endif()
+        # And push it back to the cache
+        set(HIP_ROOT_DIR ${HIP_ROOT_DIR} CACHE PATH "HIP installed location" FORCE)
+    endif()
+
+    # Find HIPCC executable
+    find_program(
+        HIP_HIPCC_EXECUTABLE
+        NAMES hipcc
+        PATHS
+        "${HIP_ROOT_DIR}"
+        ENV ROCM_PATH
+        ENV HIP_PATH
+        /opt/rocm
+        /opt/rocm/hip
+        PATH_SUFFIXES bin
+        NO_DEFAULT_PATH
+        )
+    if(NOT HIP_HIPCC_EXECUTABLE)
+        # Now search in default paths
+        find_program(HIP_HIPCC_EXECUTABLE hipcc)
+    endif()
+    mark_as_advanced(HIP_HIPCC_EXECUTABLE)
+
+    # Find HIPCONFIG executable
+    find_program(
+        HIP_HIPCONFIG_EXECUTABLE
+        NAMES hipconfig
+        PATHS
+        "${HIP_ROOT_DIR}"
+        ENV ROCM_PATH
+        ENV HIP_PATH
+        /opt/rocm
+        /opt/rocm/hip
+        PATH_SUFFIXES bin
+        NO_DEFAULT_PATH
+        )
+    if(NOT HIP_HIPCONFIG_EXECUTABLE)
+        # Now search in default paths
+        find_program(HIP_HIPCONFIG_EXECUTABLE hipconfig)
+    endif()
+    mark_as_advanced(HIP_HIPCONFIG_EXECUTABLE)
+
+    # Find HIPCC_CMAKE_LINKER_HELPER executable
+    find_program(
+        HIP_HIPCC_CMAKE_LINKER_HELPER
+        NAMES hipcc_cmake_linker_helper
+        PATHS
+        "${HIP_ROOT_DIR}"
+        ENV ROCM_PATH
+        ENV HIP_PATH
+        /opt/rocm
+        /opt/rocm/hip
+        PATH_SUFFIXES bin
+        NO_DEFAULT_PATH
+        )
+    if(NOT HIP_HIPCC_CMAKE_LINKER_HELPER)
+        # Now search in default paths
+        find_program(HIP_HIPCC_CMAKE_LINKER_HELPER hipcc_cmake_linker_helper)
+    endif()
+    mark_as_advanced(HIP_HIPCC_CMAKE_LINKER_HELPER)
+
+    if(HIP_HIPCONFIG_EXECUTABLE AND NOT HIP_VERSION)
+        # Compute the version
+        execute_process(
+            COMMAND ${HIP_HIPCONFIG_EXECUTABLE} --version
+            OUTPUT_VARIABLE _hip_version
+            ERROR_VARIABLE _hip_error
+            OUTPUT_STRIP_TRAILING_WHITESPACE
+            ERROR_STRIP_TRAILING_WHITESPACE
+            )
+        if(NOT _hip_error)
+            set(HIP_VERSION ${_hip_version} CACHE STRING "Version of HIP as computed from hipcc")
+        else()
+            set(HIP_VERSION "0.0.0" CACHE STRING "Version of HIP as computed by FindHIP()")
+        endif()
+        mark_as_advanced(HIP_VERSION)
+    endif()
+    if(HIP_VERSION)
+        string(REPLACE "." ";" _hip_version_list "${HIP_VERSION}")
+        list(GET _hip_version_list 0 HIP_VERSION_MAJOR)
+        list(GET _hip_version_list 1 HIP_VERSION_MINOR)
+        list(GET _hip_version_list 2 HIP_VERSION_PATCH)
+        set(HIP_VERSION_STRING "${HIP_VERSION}")
+    endif()
+
+    if(HIP_HIPCONFIG_EXECUTABLE AND NOT HIP_PLATFORM)
+        # Compute the platform
+        execute_process(
+            COMMAND ${HIP_HIPCONFIG_EXECUTABLE} --platform
+            OUTPUT_VARIABLE _hip_platform
+            OUTPUT_STRIP_TRAILING_WHITESPACE
+            )
+        set(HIP_PLATFORM ${_hip_platform} CACHE STRING "HIP platform as computed by hipconfig")
+        mark_as_advanced(HIP_PLATFORM)
+    endif()
+
+    if(HIP_HIPCONFIG_EXECUTABLE AND NOT HIP_COMPILER)
+        # Compute the compiler
+        execute_process(
+            COMMAND ${HIP_HIPCONFIG_EXECUTABLE} --compiler
+            OUTPUT_VARIABLE _hip_compiler
+            OUTPUT_STRIP_TRAILING_WHITESPACE
+            )
+        set(HIP_COMPILER ${_hip_compiler} CACHE STRING "HIP compiler as computed by hipconfig")
+        mark_as_advanced(HIP_COMPILER)
+    endif()
+
+    if(HIP_HIPCONFIG_EXECUTABLE AND NOT HIP_RUNTIME)
+        # Compute the runtime
+        execute_process(
+            COMMAND ${HIP_HIPCONFIG_EXECUTABLE} --runtime
+            OUTPUT_VARIABLE _hip_runtime
+            OUTPUT_STRIP_TRAILING_WHITESPACE
+            )
+        set(HIP_RUNTIME ${_hip_runtime} CACHE STRING "HIP runtime as computed by hipconfig")
+        mark_as_advanced(HIP_RUNTIME)
+    endif()
+endif()
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(
+    HIP
+    REQUIRED_VARS
+    HIP_ROOT_DIR
+    HIP_HIPCC_EXECUTABLE
+    HIP_HIPCONFIG_EXECUTABLE
+    HIP_PLATFORM
+    HIP_COMPILER
+    HIP_RUNTIME
+    VERSION_VAR HIP_VERSION
+    )
+
+###############################################################################
+# Set HIP CMAKE Flags
+###############################################################################
+# Copy the invocation styles from CXX to HIP
+set(CMAKE_HIP_ARCHIVE_CREATE ${CMAKE_CXX_ARCHIVE_CREATE})
+set(CMAKE_HIP_ARCHIVE_APPEND ${CMAKE_CXX_ARCHIVE_APPEND})
+set(CMAKE_HIP_ARCHIVE_FINISH ${CMAKE_CXX_ARCHIVE_FINISH})
+set(CMAKE_SHARED_LIBRARY_SONAME_HIP_FLAG ${CMAKE_SHARED_LIBRARY_SONAME_CXX_FLAG})
+set(CMAKE_SHARED_LIBRARY_CREATE_HIP_FLAGS ${CMAKE_SHARED_LIBRARY_CREATE_CXX_FLAGS})
+set(CMAKE_SHARED_LIBRARY_HIP_FLAGS ${CMAKE_SHARED_LIBRARY_CXX_FLAGS})
+#set(CMAKE_SHARED_LIBRARY_LINK_HIP_FLAGS ${CMAKE_SHARED_LIBRARY_LINK_CXX_FLAGS})
+set(CMAKE_SHARED_LIBRARY_RUNTIME_HIP_FLAG ${CMAKE_SHARED_LIBRARY_RUNTIME_CXX_FLAG})
+set(CMAKE_SHARED_LIBRARY_RUNTIME_HIP_FLAG_SEP ${CMAKE_SHARED_LIBRARY_RUNTIME_CXX_FLAG_SEP})
+set(CMAKE_SHARED_LIBRARY_LINK_STATIC_HIP_FLAGS ${CMAKE_SHARED_LIBRARY_LINK_STATIC_CXX_FLAGS})
+set(CMAKE_SHARED_LIBRARY_LINK_DYNAMIC_HIP_FLAGS ${CMAKE_SHARED_LIBRARY_LINK_DYNAMIC_CXX_FLAGS})
+
+set(HIP_CLANG_PARALLEL_BUILD_COMPILE_OPTIONS "")
+set(HIP_CLANG_PARALLEL_BUILD_LINK_OPTIONS "")
+
+if("${HIP_COMPILER}" STREQUAL "nvcc")
+    # Set the CMake Flags to use the nvcc Compiler.
+    set(CMAKE_HIP_CREATE_SHARED_LIBRARY "${HIP_HIPCC_CMAKE_LINKER_HELPER} <CMAKE_SHARED_LIBRARY_CXX_FLAGS> <LANGUAGE_COMPILE_FLAGS> <LINK_FLAGS> <CMAKE_SHARED_LIBRARY_CREATE_CXX_FLAGS> <SONAME_FLAG><TARGET_SONAME> -o <TARGET> <OBJECTS> <LINK_LIBRARIES>")
+    set(CMAKE_HIP_CREATE_SHARED_MODULE "${HIP_HIPCC_CMAKE_LINKER_HELPER} <CMAKE_CXX_LINK_FLAGS> <LINK_FLAGS> <OBJECTS> <SONAME_FLAG><TARGET_SONAME> -o <TARGET> <LINK_LIBRARIES> -shared" )
+    set(CMAKE_HIP_LINK_EXECUTABLE "${HIP_HIPCC_CMAKE_LINKER_HELPER} <FLAGS> <CMAKE_CXX_LINK_FLAGS> <LINK_FLAGS> <OBJECTS> -o <TARGET> <LINK_LIBRARIES>")
+elseif("${HIP_COMPILER}" STREQUAL "hcc")
+    # Set the CMake Flags to use the hcc Compiler.
+    set(CMAKE_HIP_CREATE_SHARED_LIBRARY "${HIP_HIPCC_CMAKE_LINKER_HELPER} ${HCC_HOME} <CMAKE_SHARED_LIBRARY_CXX_FLAGS> <LANGUAGE_COMPILE_FLAGS> <LINK_FLAGS> <CMAKE_SHARED_LIBRARY_CREATE_CXX_FLAGS> <SONAME_FLAG><TARGET_SONAME> -o <TARGET> <OBJECTS> <LINK_LIBRARIES>")
+    set(CMAKE_HIP_CREATE_SHARED_MODULE "${HIP_HIPCC_CMAKE_LINKER_HELPER} ${HCC_HOME} <CMAKE_CXX_LINK_FLAGS> <LINK_FLAGS> <OBJECTS> <SONAME_FLAG><TARGET_SONAME> -o <TARGET> <LINK_LIBRARIES> -shared" )
+    set(CMAKE_HIP_LINK_EXECUTABLE "${HIP_HIPCC_CMAKE_LINKER_HELPER} ${HCC_HOME} <FLAGS> <CMAKE_CXX_LINK_FLAGS> <LINK_FLAGS> <OBJECTS> -o <TARGET> <LINK_LIBRARIES>")
+elseif("${HIP_COMPILER}" STREQUAL "clang")
+    #Number of parallel jobs by default is 1
+    if(NOT DEFINED HIP_CLANG_NUM_PARALLEL_JOBS)
+      set(HIP_CLANG_NUM_PARALLEL_JOBS 1)
+    endif()
+    #Add support for parallel build and link
+    if(${CMAKE_CXX_COMPILER_ID} STREQUAL "Clang")
+      check_cxx_compiler_flag("-parallel-jobs=1" HIP_CLANG_SUPPORTS_PARALLEL_JOBS)
+    endif()
+    if(HIP_CLANG_NUM_PARALLEL_JOBS GREATER 1)
+      if(${HIP_CLANG_SUPPORTS_PARALLEL_JOBS})
+        set(HIP_CLANG_PARALLEL_BUILD_COMPILE_OPTIONS "-Wno-format-nonliteral -parallel-jobs=${HIP_CLANG_NUM_PARALLEL_JOBS}")
+        set(HIP_CLANG_PARALLEL_BUILD_LINK_OPTIONS "-parallel-jobs=${HIP_CLANG_NUM_PARALLEL_JOBS}")
+      else()
+        message("clang compiler doesn't support parallel jobs")
+      endif()
+    endif()
+
+    # Set the CMake Flags to use the HIP-Clang Compiler.
+    set(CMAKE_HIP_CREATE_SHARED_LIBRARY "${HIP_HIPCC_CMAKE_LINKER_HELPER} ${HIP_CLANG_PATH} ${HIP_CLANG_PARALLEL_BUILD_LINK_OPTIONS} <CMAKE_SHARED_LIBRARY_CXX_FLAGS> <LANGUAGE_COMPILE_FLAGS> <LINK_FLAGS> <CMAKE_SHARED_LIBRARY_CREATE_CXX_FLAGS> <SONAME_FLAG><TARGET_SONAME> -o <TARGET> <OBJECTS> <LINK_LIBRARIES>")
+    set(CMAKE_HIP_CREATE_SHARED_MODULE "${HIP_HIPCC_CMAKE_LINKER_HELPER} ${HIP_CLANG_PATH} ${HIP_CLANG_PARALLEL_BUILD_LINK_OPTIONS} <CMAKE_CXX_LINK_FLAGS> <LINK_FLAGS> <OBJECTS> <SONAME_FLAG><TARGET_SONAME> -o <TARGET> <LINK_LIBRARIES> -shared" )
+    set(CMAKE_HIP_LINK_EXECUTABLE "${HIP_HIPCC_CMAKE_LINKER_HELPER} ${HIP_CLANG_PATH} ${HIP_CLANG_PARALLEL_BUILD_LINK_OPTIONS} <FLAGS> <CMAKE_CXX_LINK_FLAGS> <LINK_FLAGS> <OBJECTS> -o <TARGET> <LINK_LIBRARIES>")
+
+    if("${HIP_RUNTIME}" STREQUAL "rocclr")
+      if(TARGET host)
+        message(STATUS "host interface - found")
+        set(HIP_HOST_INTERFACE host)
+      endif()
+    endif()
+endif()
+
+###############################################################################
+# MACRO: Locate helper files
+###############################################################################
+macro(HIP_FIND_HELPER_FILE _name _extension)
+    set(_hip_full_name "${_name}.${_extension}")
+    get_filename_component(CMAKE_CURRENT_LIST_DIR "${CMAKE_CURRENT_LIST_FILE}" PATH)
+    set(HIP_${_name} "${CMAKE_CURRENT_LIST_DIR}/FindHIP/${_hip_full_name}")
+    if(NOT EXISTS "${HIP_${_name}}")
+        set(error_message "${_hip_full_name} not found in ${CMAKE_CURRENT_LIST_DIR}/FindHIP")
+        if(HIP_FIND_REQUIRED)
+            message(FATAL_ERROR "${error_message}")
+        else()
+            if(NOT HIP_FIND_QUIETLY)
+                message(STATUS "${error_message}")
+            endif()
+        endif()
+    endif()
+    # Set this variable as internal, so the user isn't bugged with it.
+    set(HIP_${_name} ${HIP_${_name}} CACHE INTERNAL "Location of ${_full_name}" FORCE)
+endmacro()
+
+###############################################################################
+hip_find_helper_file(run_make2cmake cmake)
+hip_find_helper_file(run_hipcc cmake)
+###############################################################################
+
+###############################################################################
+# MACRO: Reset compiler flags
+###############################################################################
+macro(HIP_RESET_FLAGS)
+    unset(HIP_HIPCC_FLAGS)
+    unset(HIP_HCC_FLAGS)
+    unset(HIP_CLANG_FLAGS)
+    unset(HIP_NVCC_FLAGS)
+    foreach(config ${_hip_configuration_types})
+        string(TOUPPER ${config} config_upper)
+        unset(HIP_HIPCC_FLAGS_${config_upper})
+        unset(HIP_HCC_FLAGS_${config_upper})
+        unset(HIP_CLANG_FLAGS_${config_upper})
+        unset(HIP_NVCC_FLAGS_${config_upper})
+    endforeach()
+endmacro()
+
+###############################################################################
+# MACRO: Separate the options from the sources
+###############################################################################
+macro(HIP_GET_SOURCES_AND_OPTIONS _sources _cmake_options _hipcc_options _hcc_options _clang_options _nvcc_options)
+    set(${_sources})
+    set(${_cmake_options})
+    set(${_hipcc_options})
+    set(${_hcc_options})
+    set(${_clang_options})
+    set(${_nvcc_options})
+    set(_hipcc_found_options FALSE)
+    set(_hcc_found_options FALSE)
+    set(_clang_found_options FALSE)
+    set(_nvcc_found_options FALSE)
+    foreach(arg ${ARGN})
+        if("x${arg}" STREQUAL "xHIPCC_OPTIONS")
+            set(_hipcc_found_options TRUE)
+            set(_hcc_found_options FALSE)
+            set(_clang_found_options FALSE)
+            set(_nvcc_found_options FALSE)
+        elseif("x${arg}" STREQUAL "xHCC_OPTIONS")
+            set(_hipcc_found_options FALSE)
+            set(_hcc_found_options TRUE)
+            set(_clang_found_options FALSE)
+            set(_nvcc_found_options FALSE)
+        elseif("x${arg}" STREQUAL "xCLANG_OPTIONS")
+            set(_hipcc_found_options FALSE)
+            set(_hcc_found_options FALSE)
+            set(_clang_found_options TRUE)
+            set(_nvcc_found_options FALSE)
+        elseif("x${arg}" STREQUAL "xNVCC_OPTIONS")
+            set(_hipcc_found_options FALSE)
+            set(_hcc_found_options FALSE)
+            set(_clang_found_options FALSE)
+            set(_nvcc_found_options TRUE)
+        elseif(
+                "x${arg}" STREQUAL "xEXCLUDE_FROM_ALL" OR
+                "x${arg}" STREQUAL "xSTATIC" OR
+                "x${arg}" STREQUAL "xSHARED" OR
+                "x${arg}" STREQUAL "xMODULE"
+                )
+            list(APPEND ${_cmake_options} ${arg})
+        else()
+            if(_hipcc_found_options)
+                list(APPEND ${_hipcc_options} ${arg})
+            elseif(_hcc_found_options)
+                list(APPEND ${_hcc_options} ${arg})
+            elseif(_clang_found_options)
+                list(APPEND ${_clang_options} ${arg})
+            elseif(_nvcc_found_options)
+                list(APPEND ${_nvcc_options} ${arg})
+            else()
+                # Assume this is a file
+                list(APPEND ${_sources} ${arg})
+            endif()
+        endif()
+    endforeach()
+endmacro()
+
+###############################################################################
+# MACRO: Add include directories to pass to the hipcc command
+###############################################################################
+set(HIP_HIPCC_INCLUDE_ARGS_USER "")
+macro(HIP_INCLUDE_DIRECTORIES)
+    foreach(dir ${ARGN})
+        list(APPEND HIP_HIPCC_INCLUDE_ARGS_USER $<$<BOOL:${dir}>:-I${dir}>)
+    endforeach()
+endmacro()
+
+###############################################################################
+# FUNCTION: Helper to avoid clashes of files with the same basename but different paths
+###############################################################################
+function(HIP_COMPUTE_BUILD_PATH path build_path)
+    # Convert to cmake style paths
+    file(TO_CMAKE_PATH "${path}" bpath)
+    if(IS_ABSOLUTE "${bpath}")
+        string(FIND "${bpath}" "${CMAKE_CURRENT_BINARY_DIR}" _binary_dir_pos)
+        if(_binary_dir_pos EQUAL 0)
+            file(RELATIVE_PATH bpath "${CMAKE_CURRENT_BINARY_DIR}" "${bpath}")
+        else()
+            file(RELATIVE_PATH bpath "${CMAKE_CURRENT_SOURCE_DIR}" "${bpath}")
+        endif()
+    endif()
+
+    # Remove leading /
+    string(REGEX REPLACE "^[/]+" "" bpath "${bpath}")
+    # Avoid absolute paths by removing ':'
+    string(REPLACE ":" "_" bpath "${bpath}")
+    # Avoid relative paths that go up the tree
+    string(REPLACE "../" "__/" bpath "${bpath}")
+    # Avoid spaces
+    string(REPLACE " " "_" bpath "${bpath}")
+    # Strip off the filename
+    get_filename_component(bpath "${bpath}" PATH)
+
+    set(${build_path} "${bpath}" PARENT_SCOPE)
+endfunction()
+
+###############################################################################
+# MACRO: Parse OPTIONS from ARGN & set variables prefixed by _option_prefix
+###############################################################################
+macro(HIP_PARSE_HIPCC_OPTIONS _option_prefix)
+    set(_hip_found_config)
+    foreach(arg ${ARGN})
+        # Determine if we are dealing with a per-configuration flag
+        foreach(config ${_hip_configuration_types})
+            string(TOUPPER ${config} config_upper)
+            if(arg STREQUAL "${config_upper}")
+                set(_hip_found_config _${arg})
+                # Clear arg to prevent it from being processed anymore
+                set(arg)
+            endif()
+        endforeach()
+        if(arg)
+            list(APPEND ${_option_prefix}${_hip_found_config} "${arg}")
+        endif()
+    endforeach()
+endmacro()
+
+###############################################################################
+# MACRO: Try and include dependency file if it exists
+###############################################################################
+macro(HIP_INCLUDE_HIPCC_DEPENDENCIES dependency_file)
+    set(HIP_HIPCC_DEPEND)
+    set(HIP_HIPCC_DEPEND_REGENERATE FALSE)
+
+    # Create the dependency file if it doesn't exist
+    if(NOT EXISTS ${dependency_file})
+        file(WRITE ${dependency_file} "# Generated by: FindHIP.cmake. Do not edit.\n")
+    endif()
+    # Include the dependency file
+    include(${dependency_file})
+
+    # Verify the existence of all the included files
+    if(HIP_HIPCC_DEPEND)
+        foreach(f ${HIP_HIPCC_DEPEND})
+            if(NOT EXISTS ${f})
+                # If they aren't there, regenerate the file again
+                set(HIP_HIPCC_DEPEND_REGENERATE TRUE)
+            endif()
+        endforeach()
+    else()
+        # No dependencies, so regenerate the file
+        set(HIP_HIPCC_DEPEND_REGENERATE TRUE)
+    endif()
+
+    # Regenerate the dependency file if needed
+    if(HIP_HIPCC_DEPEND_REGENERATE)
+        set(HIP_HIPCC_DEPEND ${dependency_file})
+        file(WRITE ${dependency_file} "# Generated by: FindHIP.cmake. Do not edit.\n")
+    endif()
+endmacro()
+
+###############################################################################
+# MACRO: Prepare cmake commands for the target
+###############################################################################
+macro(HIP_PREPARE_TARGET_COMMANDS _target _format _generated_files _source_files)
+    set(_hip_flags "")
+    string(TOUPPER "${CMAKE_BUILD_TYPE}" _hip_build_configuration)
+    if(HIP_HOST_COMPILATION_CPP)
+        set(HIP_C_OR_CXX CXX)
+    else()
+        set(HIP_C_OR_CXX C)
+    endif()
+    set(generated_extension ${CMAKE_${HIP_C_OR_CXX}_OUTPUT_EXTENSION})
+
+    # Initialize list of includes with those specified by the user. Append with
+    # ones specified to cmake directly.
+    set(HIP_HIPCC_INCLUDE_ARGS ${HIP_HIPCC_INCLUDE_ARGS_USER})
+
+    # Add the include directories
+    set(include_directories_generator "$<TARGET_PROPERTY:${_target},INCLUDE_DIRECTORIES>")
+    list(APPEND HIP_HIPCC_INCLUDE_ARGS "$<$<BOOL:${include_directories_generator}>:-I$<JOIN:${include_directories_generator}, -I>>")
+
+    get_directory_property(_hip_include_directories INCLUDE_DIRECTORIES)
+    list(REMOVE_DUPLICATES _hip_include_directories)
+    if(_hip_include_directories)
+        foreach(dir ${_hip_include_directories})
+            list(APPEND HIP_HIPCC_INCLUDE_ARGS $<$<BOOL:${dir}>:-I${dir}>)
+        endforeach()
+    endif()
+
+    HIP_GET_SOURCES_AND_OPTIONS(_hip_sources _hip_cmake_options _hipcc_options _hcc_options _clang_options _nvcc_options ${ARGN})
+    HIP_PARSE_HIPCC_OPTIONS(HIP_HIPCC_FLAGS ${_hipcc_options})
+    HIP_PARSE_HIPCC_OPTIONS(HIP_HCC_FLAGS ${_hcc_options})
+    HIP_PARSE_HIPCC_OPTIONS(HIP_CLANG_FLAGS ${_clang_options})
+    HIP_PARSE_HIPCC_OPTIONS(HIP_NVCC_FLAGS ${_nvcc_options})
+
+    # Add the compile definitions
+    set(compile_definition_generator "$<TARGET_PROPERTY:${_target},COMPILE_DEFINITIONS>")
+    list(APPEND HIP_HIPCC_FLAGS "$<$<BOOL:${compile_definition_generator}>:-D$<JOIN:${compile_definition_generator}, -D>>")
+
+    # Check if we are building shared library.
+    set(_hip_build_shared_libs FALSE)
+    list(FIND _hip_cmake_options SHARED _hip_found_SHARED)
+    list(FIND _hip_cmake_options MODULE _hip_found_MODULE)
+    if(_hip_found_SHARED GREATER -1 OR _hip_found_MODULE GREATER -1)
+        set(_hip_build_shared_libs TRUE)
+    endif()
+    list(FIND _hip_cmake_options STATIC _hip_found_STATIC)
+    if(_hip_found_STATIC GREATER -1)
+        set(_hip_build_shared_libs FALSE)
+    endif()
+
+    # If we are building a shared library, add extra flags to HIP_HIPCC_FLAGS
+    if(_hip_build_shared_libs)
+        list(APPEND HIP_HCC_FLAGS "-fPIC")
+        list(APPEND HIP_CLANG_FLAGS "-fPIC")
+        list(APPEND HIP_NVCC_FLAGS "--shared -Xcompiler '-fPIC'")
+    endif()
+
+    # Set host compiler
+    set(HIP_HOST_COMPILER "${CMAKE_${HIP_C_OR_CXX}_COMPILER}")
+
+    # Set compiler flags
+    set(_HIP_HOST_FLAGS "set(CMAKE_HOST_FLAGS ${CMAKE_${HIP_C_OR_CXX}_FLAGS})")
+    set(_HIP_HIPCC_FLAGS "set(HIP_HIPCC_FLAGS ${HIP_HIPCC_FLAGS})")
+    set(_HIP_HCC_FLAGS "set(HIP_HCC_FLAGS ${HIP_HCC_FLAGS})")
+    set(_HIP_CLANG_FLAGS "set(HIP_CLANG_FLAGS ${HIP_CLANG_FLAGS})")
+    set(_HIP_NVCC_FLAGS "set(HIP_NVCC_FLAGS ${HIP_NVCC_FLAGS})")
+    foreach(config ${_hip_configuration_types})
+        string(TOUPPER ${config} config_upper)
+        set(_HIP_HOST_FLAGS "${_HIP_HOST_FLAGS}\nset(CMAKE_HOST_FLAGS_${config_upper} ${CMAKE_${HIP_C_OR_CXX}_FLAGS_${config_upper}})")
+        set(_HIP_HIPCC_FLAGS "${_HIP_HIPCC_FLAGS}\nset(HIP_HIPCC_FLAGS_${config_upper} ${HIP_HIPCC_FLAGS_${config_upper}})")
+        set(_HIP_HCC_FLAGS "${_HIP_HCC_FLAGS}\nset(HIP_HCC_FLAGS_${config_upper} ${HIP_HCC_FLAGS_${config_upper}})")
+        set(_HIP_CLANG_FLAGS "${_HIP_CLANG_FLAGS}\nset(HIP_CLANG_FLAGS_${config_upper} ${HIP_CLANG_FLAGS_${config_upper}})")
+        set(_HIP_NVCC_FLAGS "${_HIP_NVCC_FLAGS}\nset(HIP_NVCC_FLAGS_${config_upper} ${HIP_NVCC_FLAGS_${config_upper}})")
+    endforeach()
+
+    # Reset the output variable
+    set(_hip_generated_files "")
+    set(_hip_source_files "")
+
+    # Iterate over all arguments and create custom commands for all source files
+    foreach(file ${ARGN})
+        # Ignore any file marked as a HEADER_FILE_ONLY
+        get_source_file_property(_is_header ${file} HEADER_FILE_ONLY)
+        # Allow per source file overrides of the format. Also allows compiling non .cu files.
+        get_source_file_property(_hip_source_format ${file} HIP_SOURCE_PROPERTY_FORMAT)
+        if((${file} MATCHES "\\.cu$" OR _hip_source_format) AND NOT _is_header)
+            set(host_flag FALSE)
+        else()
+            set(host_flag TRUE)
+        endif()
+
+        if(NOT host_flag)
+            # Determine output directory
+            HIP_COMPUTE_BUILD_PATH("${file}" hip_build_path)
+            set(hip_compile_output_dir "${CMAKE_CURRENT_BINARY_DIR}/CMakeFiles/${_target}.dir/${hip_build_path}")
+
+            get_filename_component(basename ${file} NAME)
+            set(generated_file_path "${hip_compile_output_dir}/${CMAKE_CFG_INTDIR}")
+            set(generated_file_basename "${_target}_generated_${basename}${generated_extension}")
+
+            # Set file names
+            set(generated_file "${generated_file_path}/${generated_file_basename}")
+            set(cmake_dependency_file "${hip_compile_output_dir}/${generated_file_basename}.depend")
+            set(custom_target_script_pregen "${hip_compile_output_dir}/${generated_file_basename}.cmake.pre-gen")
+            set(custom_target_script "${hip_compile_output_dir}/${generated_file_basename}.cmake")
+
+            # Set properties for object files
+            set_source_files_properties("${generated_file}"
+                PROPERTIES
+                EXTERNAL_OBJECT true # This is an object file not to be compiled, but only be linked
+                )
+
+            # Don't add CMAKE_CURRENT_SOURCE_DIR if the path is already an absolute path
+            get_filename_component(file_path "${file}" PATH)
+            if(IS_ABSOLUTE "${file_path}")
+                set(source_file "${file}")
+            else()
+                set(source_file "${CMAKE_CURRENT_SOURCE_DIR}/${file}")
+            endif()
+
+            # Bring in the dependencies
+            HIP_INCLUDE_HIPCC_DEPENDENCIES(${cmake_dependency_file})
+
+            # Configure the build script
+            configure_file("${HIP_run_hipcc}" "${custom_target_script_pregen}" @ONLY)
+            file(GENERATE
+                OUTPUT "${custom_target_script}"
+                INPUT "${custom_target_script_pregen}"
+                )
+            set(main_dep DEPENDS ${source_file})
+            if(CMAKE_GENERATOR MATCHES "Makefiles")
+                set(verbose_output "$(VERBOSE)")
+            elseif(HIP_VERBOSE_BUILD)
+                set(verbose_output ON)
+            else()
+                set(verbose_output OFF)
+            endif()
+
+            # Create up the comment string
+            file(RELATIVE_PATH generated_file_relative_path "${CMAKE_BINARY_DIR}" "${generated_file}")
+            set(hip_build_comment_string "Building HIPCC object ${generated_file_relative_path}")
+
+            # Build the generated file and dependency file
+            add_custom_command(
+                OUTPUT ${generated_file}
+                # These output files depend on the source_file and the contents of cmake_dependency_file
+                ${main_dep}
+                DEPENDS ${HIP_HIPCC_DEPEND}
+                DEPENDS ${custom_target_script}
+                # Make sure the output directory exists before trying to write to it.
+                COMMAND ${CMAKE_COMMAND} -E make_directory "${generated_file_path}"
+                COMMAND ${CMAKE_COMMAND} ARGS
+                -D verbose:BOOL=${verbose_output}
+                -D build_configuration:STRING=${_hip_build_configuration}
+                -D "generated_file:STRING=${generated_file}"
+                -P "${custom_target_script}"
+                WORKING_DIRECTORY "${hip_compile_output_dir}"
+                COMMENT "${hip_build_comment_string}"
+                )
+
+            # Make sure the build system knows the file is generated
+            set_source_files_properties(${generated_file} PROPERTIES GENERATED TRUE)
+            list(APPEND _hip_generated_files ${generated_file})
+            list(APPEND _hip_source_files ${file})
+        endif()
+    endforeach()
+
+    # Set the return parameter
+    set(${_generated_files} ${_hip_generated_files})
+    set(${_source_files} ${_hip_source_files})
+endmacro()
+
+###############################################################################
+# HIP_ADD_EXECUTABLE
+###############################################################################
+macro(HIP_ADD_EXECUTABLE hip_target)
+    # Separate the sources from the options
+    HIP_GET_SOURCES_AND_OPTIONS(_sources _cmake_options _hipcc_options _hcc_options _clang_options _nvcc_options ${ARGN})
+    HIP_PREPARE_TARGET_COMMANDS(${hip_target} OBJ _generated_files _source_files ${_sources} HIPCC_OPTIONS ${_hipcc_options} HCC_OPTIONS ${_hcc_options} CLANG_OPTIONS ${_clang_options} NVCC_OPTIONS ${_nvcc_options})
+    if(_source_files)
+        list(REMOVE_ITEM _sources ${_source_files})
+    endif()
+    if("${HIP_COMPILER}" STREQUAL "hcc")
+        if("x${HCC_HOME}" STREQUAL "x")
+            if (DEFINED ENV{ROCM_PATH})
+                set(HCC_HOME "$ENV{ROCM_PATH}/hcc")
+            elseif(DEFINED ENV{HIP_PATH})
+                set(HCC_HOME "$ENV{HIP_PATH}/../hcc")
+            else()
+                set(HCC_HOME "/opt/rocm/hcc")
+            endif()
+        endif()
+        set(CMAKE_HIP_LINK_EXECUTABLE "${HIP_HIPCC_CMAKE_LINKER_HELPER} ${HCC_HOME} <FLAGS> <CMAKE_CXX_LINK_FLAGS> <LINK_FLAGS> <OBJECTS> -o <TARGET> <LINK_LIBRARIES>")
+    elseif("${HIP_COMPILER}" STREQUAL "clang")
+        if("x${HIP_CLANG_PATH}" STREQUAL "x")
+            if(DEFINED ENV{HIP_CLANG_PATH})
+                set(HIP_CLANG_PATH $ENV{HIP_CLANG_PATH})
+            elseif(DEFINED ENV{ROCM_PATH})
+                set(HIP_CLANG_PATH "$ENV{ROCM_PATH}/llvm/bin")
+            elseif(DEFINED ENV{HIP_PATH})
+                set(HIP_CLANG_PATH "$ENV{HIP_PATH}/../llvm/bin")
+            else()
+                set(HIP_CLANG_PATH "/opt/rocm/llvm/bin")
+            endif()
+        endif()
+        set(CMAKE_HIP_LINK_EXECUTABLE "${HIP_HIPCC_CMAKE_LINKER_HELPER} ${HIP_CLANG_PATH} ${HIP_CLANG_PARALLEL_BUILD_LINK_OPTIONS} <FLAGS> <CMAKE_CXX_LINK_FLAGS> <LINK_FLAGS> <OBJECTS> -o <TARGET> <LINK_LIBRARIES>")
+    else()
+        set(CMAKE_HIP_LINK_EXECUTABLE "${HIP_HIPCC_CMAKE_LINKER_HELPER} <FLAGS> <CMAKE_CXX_LINK_FLAGS> <LINK_FLAGS> <OBJECTS> -o <TARGET> <LINK_LIBRARIES>")
+    endif()
+    if ("${_sources}" STREQUAL "")
+        add_executable(${hip_target} ${_cmake_options} ${_generated_files} "")
+    else()
+        add_executable(${hip_target} ${_cmake_options} ${_generated_files} ${_sources})
+    endif()
+    set_target_properties(${hip_target} PROPERTIES LINKER_LANGUAGE HIP)
+    # Link with host
+    if (HIP_HOST_INTERFACE)
+        # hip rt should be rocclr, compiler should be clang
+        target_link_libraries(${hip_target} ${HIP_HOST_INTERFACE})
+    endif()
+endmacro()
+
+###############################################################################
+# HIP_ADD_LIBRARY
+###############################################################################
+macro(HIP_ADD_LIBRARY hip_target)
+    # Separate the sources from the options
+    HIP_GET_SOURCES_AND_OPTIONS(_sources _cmake_options _hipcc_options _hcc_options _clang_options _nvcc_options ${ARGN})
+    HIP_PREPARE_TARGET_COMMANDS(${hip_target} OBJ _generated_files _source_files ${_sources} ${_cmake_options} HIPCC_OPTIONS ${_hipcc_options} HCC_OPTIONS ${_hcc_options} CLANG_OPTIONS ${_clang_options} NVCC_OPTIONS ${_nvcc_options})
+    if(_source_files)
+        list(REMOVE_ITEM _sources ${_source_files})
+    endif()
+    if ("${_sources}" STREQUAL "")
+        add_library(${hip_target} ${_cmake_options} ${_generated_files} "")
+    else()
+        add_library(${hip_target} ${_cmake_options} ${_generated_files} ${_sources})
+    endif()
+    set_target_properties(${hip_target} PROPERTIES LINKER_LANGUAGE ${HIP_C_OR_CXX})
+    # Link with host
+    if (HIP_HOST_INTERFACE)
+        # hip rt should be rocclr, compiler should be clang
+        target_link_libraries(${hip_target} ${HIP_HOST_INTERFACE})
+    endif()
+endmacro()
+
+# vim: ts=4:sw=4:expandtab:smartindent
@@ -0,0 +1,182 @@
+###############################################################################
+# Runs commands using HIPCC
+###############################################################################
+
+###############################################################################
+# This file runs the hipcc commands to produce the desired output file
+# along with the dependency file needed by CMake to compute dependencies.
+#
+# Input variables:
+#
+# verbose:BOOL=<>               OFF: Be as quiet as possible (default)
+#                               ON : Describe each step
+# build_configuration:STRING=<> Build configuration. Defaults to Debug.
+# generated_file:STRING=<>      File to generate. Mandatory argument.
+
+if(NOT build_configuration)
+    set(build_configuration Debug)
+endif()
+if(NOT generated_file)
+    message(FATAL_ERROR "You must specify generated_file on the command line")
+endif()
+
+# Set these up as variables to make reading the generated file easier
+set(HIP_HIPCC_EXECUTABLE "@HIP_HIPCC_EXECUTABLE@") # path
+set(HIP_HIPCONFIG_EXECUTABLE "@HIP_HIPCONFIG_EXECUTABLE@") #path
+set(HIP_HOST_COMPILER "@HIP_HOST_COMPILER@") # path
+set(CMAKE_COMMAND "@CMAKE_COMMAND@") # path
+set(HIP_run_make2cmake "@HIP_run_make2cmake@") # path
+set(HCC_HOME "@HCC_HOME@") #path
+set(HIP_CLANG_PATH "@HIP_CLANG_PATH@") #path
+set(HIP_CLANG_PARALLEL_BUILD_COMPILE_OPTIONS "@HIP_CLANG_PARALLEL_BUILD_COMPILE_OPTIONS@")
+
+@HIP_HOST_FLAGS@
+@_HIP_HIPCC_FLAGS@
+@_HIP_HCC_FLAGS@
+@_HIP_CLANG_FLAGS@
+@_HIP_NVCC_FLAGS@
+#Needed to bring the HIP_HIPCC_INCLUDE_ARGS variable in scope
+set(HIP_HIPCC_INCLUDE_ARGS @HIP_HIPCC_INCLUDE_ARGS@) # list
+
+set(cmake_dependency_file "@cmake_dependency_file@") # path
+set(source_file "@source_file@") # path
+set(host_flag "@host_flag@") # bool
+
+# Determine compiler and compiler flags
+execute_process(COMMAND ${HIP_HIPCONFIG_EXECUTABLE} --platform OUTPUT_VARIABLE HIP_PLATFORM OUTPUT_STRIP_TRAILING_WHITESPACE)
+execute_process(COMMAND ${HIP_HIPCONFIG_EXECUTABLE} --compiler OUTPUT_VARIABLE HIP_COMPILER OUTPUT_STRIP_TRAILING_WHITESPACE)
+execute_process(COMMAND ${HIP_HIPCONFIG_EXECUTABLE} --runtime OUTPUT_VARIABLE HIP_RUNTIME OUTPUT_STRIP_TRAILING_WHITESPACE)
+if(NOT host_flag)
+    set(__CC ${HIP_HIPCC_EXECUTABLE})
+    if("${HIP_PLATFORM}" STREQUAL "amd")
+        if("${HIP_COMPILER}" STREQUAL "hcc")
+            if(NOT "x${HCC_HOME}" STREQUAL "x")
+                set(ENV{HCC_HOME} ${HCC_HOME})
+            endif()
+            set(__CC_FLAGS ${HIP_HIPCC_FLAGS} ${HIP_HCC_FLAGS} ${HIP_HIPCC_FLAGS_${build_configuration}} ${HIP_HCC_FLAGS_${build_configuration}})
+        elseif("${HIP_COMPILER}" STREQUAL "clang")
+            if(NOT "x${HIP_CLANG_PATH}" STREQUAL "x")
+                set(ENV{HIP_CLANG_PATH} ${HIP_CLANG_PATH})
+            endif()
+            # Temporarily include HIP_HCC_FLAGS for HIP-Clang for PyTorch builds
+            set(__CC_FLAGS ${HIP_CLANG_PARALLEL_BUILD_COMPILE_OPTIONS} ${HIP_HIPCC_FLAGS} ${HIP_HCC_FLAGS} ${HIP_CLANG_FLAGS} ${HIP_HIPCC_FLAGS_${build_configuration}} ${HIP_HCC_FLAGS_${build_configuration}} ${HIP_CLANG_FLAGS_${build_configuration}})
+        endif()
+    else()
+        set(__CC_FLAGS ${HIP_HIPCC_FLAGS} ${HIP_NVCC_FLAGS} ${HIP_HIPCC_FLAGS_${build_configuration}} ${HIP_NVCC_FLAGS_${build_configuration}})
+    endif()
+else()
+    set(__CC ${HIP_HOST_COMPILER})
+    set(__CC_FLAGS ${CMAKE_HOST_FLAGS} ${CMAKE_HOST_FLAGS_${build_configuration}})
+endif()
+set(__CC_INCLUDES ${HIP_HIPCC_INCLUDE_ARGS})
+
+# hip_execute_process - Executes a command with optional command echo and status message.
+#   status     - Status message to print if verbose is true
+#   command    - COMMAND argument from the usual execute_process argument structure
+#   ARGN       - Remaining arguments are the command with arguments
+#   HIP_result - Return value from running the command
+macro(hip_execute_process status command)
+    set(_command ${command})
+    if(NOT "x${_command}" STREQUAL "xCOMMAND")
+        message(FATAL_ERROR "Malformed call to hip_execute_process.  Missing COMMAND as second argument. (command = ${command})")
+    endif()
+    if(verbose)
+        execute_process(COMMAND "${CMAKE_COMMAND}" -E echo -- ${status})
+        # Build command string to print
+        set(hip_execute_process_string)
+        foreach(arg ${ARGN})
+            # Escape quotes if any
+            string(REPLACE "\"" "\\\"" arg ${arg})
+            # Surround args with spaces with quotes
+            if(arg MATCHES " ")
+                list(APPEND hip_execute_process_string "\"${arg}\"")
+            else()
+                list(APPEND hip_execute_process_string ${arg})
+            endif()
+        endforeach()
+        # Echo the command
+        execute_process(COMMAND ${CMAKE_COMMAND} -E echo ${hip_execute_process_string})
+    endif()
+    # Run the command
+    execute_process(COMMAND ${ARGN} RESULT_VARIABLE HIP_result)
+endmacro()
+
+# Delete the target file
+hip_execute_process(
+    "Removing ${generated_file}"
+    COMMAND "${CMAKE_COMMAND}" -E remove "${generated_file}"
+    )
+
+# Generate the dependency file
+hip_execute_process(
+    "Generating dependency file: ${cmake_dependency_file}.pre"
+    COMMAND "${__CC}"
+    -M
+    "${source_file}"
+    -o "${cmake_dependency_file}.pre"
+    ${__CC_FLAGS}
+    ${__CC_INCLUDES}
+    )
+
+if(HIP_result)
+    message(FATAL_ERROR "Error generating ${generated_file}")
+endif()
+
+# Generate the cmake readable dependency file to a temp file
+hip_execute_process(
+    "Generating temporary cmake readable file: ${cmake_dependency_file}.tmp"
+    COMMAND "${CMAKE_COMMAND}"
+    -D "input_file:FILEPATH=${cmake_dependency_file}.pre"
+    -D "output_file:FILEPATH=${cmake_dependency_file}.tmp"
+    -D "verbose=${verbose}"
+    -P "${HIP_run_make2cmake}"
+    )
+
+if(HIP_result)
+    message(FATAL_ERROR "Error generating ${generated_file}")
+endif()
+
+# Copy the file if it is different
+hip_execute_process(
+    "Copy if different ${cmake_dependency_file}.tmp to ${cmake_dependency_file}"
+    COMMAND "${CMAKE_COMMAND}" -E copy_if_different "${cmake_dependency_file}.tmp" "${cmake_dependency_file}"
+    )
+
+if(HIP_result)
+    message(FATAL_ERROR "Error generating ${generated_file}")
+endif()
+
+# Delete the temporary file
+hip_execute_process(
+    "Removing ${cmake_dependency_file}.tmp and ${cmake_dependency_file}.pre"
+    COMMAND "${CMAKE_COMMAND}" -E remove "${cmake_dependency_file}.tmp" "${cmake_dependency_file}.pre"
+    )
+
+if(HIP_result)
+    message(FATAL_ERROR "Error generating ${generated_file}")
+endif()
+
+# Generate the output file
+hip_execute_process(
+    "Generating ${generated_file}"
+    COMMAND "${__CC}"
+    -c
+    "${source_file}"
+    -o "${generated_file}"
+    ${__CC_FLAGS}
+    ${__CC_INCLUDES}
+    )
+
+if(HIP_result)
+    # Make sure that we delete the output file
+    hip_execute_process(
+        "Removing ${generated_file}"
+        COMMAND "${CMAKE_COMMAND}" -E remove "${generated_file}"
+        )
+    message(FATAL_ERROR "Error generating file ${generated_file}")
+else()
+    if(verbose)
+        message("Generated ${generated_file} successfully.")
+    endif()
+endif()
+# vim: ts=4:sw=4:expandtab:smartindent
@@ -0,0 +1,50 @@
+###############################################################################
+# Computes dependencies using HIPCC
+###############################################################################
+
+###############################################################################
+# This file converts dependency files generated using hipcc to a format that
+# cmake can understand.
+
+# Input variables:
+#
+# input_file:STRING=<> Dependency file to parse. Required argument
+# output_file:STRING=<> Output file to generate. Required argument
+
+if(NOT input_file OR NOT output_file)
+    message(FATAL_ERROR "You must specify input_file and output_file on the command line")
+endif()
+
+file(READ ${input_file} depend_text)
+
+if (NOT "${depend_text}" STREQUAL "")
+    string(REPLACE " /" "\n/" depend_text ${depend_text})
+    string(REGEX REPLACE "^.*:" "" depend_text ${depend_text})
+    string(REGEX REPLACE "[ \\\\]*\n" ";" depend_text ${depend_text})
+
+    set(dependency_list "")
+
+    foreach(file ${depend_text})
+        string(REGEX REPLACE "^ +" "" file ${file})
+        if(NOT EXISTS "${file}")
+            message(WARNING " Removing non-existent dependency file: ${file}")
+            set(file "")
+        endif()
+
+        if(NOT IS_DIRECTORY "${file}")
+            get_filename_component(file_absolute "${file}" ABSOLUTE)
+            list(APPEND dependency_list "${file_absolute}")
+        endif()
+    endforeach()
+endif()
+
+# Remove the duplicate entries and sort them.
+list(REMOVE_DUPLICATES dependency_list)
+list(SORT dependency_list)
+
+foreach(file ${dependency_list})
+    set(hip_hipcc_depend "${hip_hipcc_depend} \"${file}\"\n")
+endforeach()
+
+file(WRITE ${output_file} "# Generated by: FindHIP.cmake. Do not edit.\nSET(HIP_HIPCC_DEPEND\n ${hip_hipcc_depend})\n\n")
+# vim: ts=4:sw=4:expandtab:smartindent
@@ -14,33 +14,10 @@
 #   - HYPRE_LIBRARIES
 #   - HYPRE_INCLUDE_DIRS
 #   - HYPRE_VERSION
-#   - HYPRE_USING_HIP (internal)
-
-if (HYPRE_FOUND)
-  if (HYPRE_USING_HIP)
-    find_package(rocsparse REQUIRED)
-    find_package(rocrand REQUIRED)
-  endif()
-  return()
-endif()

 include(MfemCmakeUtilities)
 mfem_find_package(HYPRE HYPRE HYPRE_DIR "include" "HYPRE.h" "lib" "HYPRE"
-  "Paths to headers required by HYPRE." "Libraries required by HYPRE."
-  CHECK_BUILD HYPRE_USING_HIP FALSE
-  "
-#undef HYPRE_USING_HIP
-#include <HYPRE_config.h>
-
-#ifndef HYPRE_USING_HIP
-#error HYPRE is built without HIP.
-#endif
-
-int main()
-{
-   return 0;
-}
-")
+  "Paths to headers required by HYPRE." "Libraries required by HYPRE.")

 if (HYPRE_FOUND AND (NOT HYPRE_VERSION))
  try_run(HYPRE_VERSION_RUN_RESULT HYPRE_VERSION_COMPILE_RESULT
@@ -56,12 +33,3 @@ if (HYPRE_FOUND AND (NOT HYPRE_VERSION))
    message(FATAL_ERROR "Unable to determine HYPRE version.")
  endif()
 endif()
-
-if (HYPRE_FOUND AND HYPRE_USING_HIP)
-  find_package(rocsparse REQUIRED)
-  find_package(rocrand REQUIRED)
-  list(APPEND HYPRE_LIBRARIES ${rocsparse_LIBRARIES} ${rocrand_LIBRARIES})
-  set(HYPRE_LIBRARIES ${HYPRE_LIBRARIES} CACHE STRING
-      "HYPRE libraries + dependencies." FORCE)
-  message(STATUS "Updated HYPRE_LIBRARIES: ${HYPRE_LIBRARIES}")
-endif()
@@ -1,19 +0,0 @@
-# Copyright (c) 2010-2021, Lawrence Livermore National Security, LLC. Produced
-# at the Lawrence Livermore National Laboratory. All Rights reserved. See files
-# LICENSE and NOTICE for details. LLNL-CODE-806117.
-#
-# This file is part of the MFEM library. For more information and source code
-# availability visit https://mfem.org.
-#
-# MFEM is free software; you can redistribute it and/or modify it under the
-# terms of the BSD-3 license. We welcome feedback and contributions, see file
-# CONTRIBUTING.md for details.
-
-# Defines the following variables:
-#   - PARELAG_FOUND
-#   - PARELAG_LIBRARIES
-#   - PARELAG_INCLUDE_DIRS
-
-include(MfemCmakeUtilities)
-mfem_find_package(PARELAG PARELAG PARELAG_DIR "" "" "" ""
-  "Paths to headers required by ParELAG." "Libraries required by ParELAG.")
@@ -46,7 +46,8 @@ endfunction()
 # Wrapper for add_executable that calls the HIP wrapper if applicable
 macro(mfem_add_executable NAME)
  if (MFEM_USE_HIP)
-    add_executable(${NAME} ${ARGN})
+    hip_add_executable(${NAME} ${ARGN})
+    set_target_properties(${NAME} PROPERTIES LINKER_LANGUAGE CXX)
  else()
    add_executable(${NAME} ${ARGN})
  endif()
@@ -55,7 +56,7 @@ endmacro()
 # Wrapper for add_library that calls the HIP wrapper if applicable
 macro(mfem_add_library NAME)
  if (MFEM_USE_HIP)
-    add_library(${NAME} ${ARGN})
+    hip_add_library(${NAME} ${ARGN})
  else()
    add_library(${NAME} ${ARGN})
  endif()
@@ -91,14 +92,14 @@ macro(add_mfem_examples EXE_SRCS)
    # If CUDA is enabled, tag source files to be compiled with nvcc.
    if (MFEM_USE_CUDA)
      set_source_files_properties(${SRC_FILE} PROPERTIES LANGUAGE CUDA)
+    elseif(MFEM_USE_HIP)
+      set_source_files_properties(${SRC_FILE} PROPERTIES HIP_SOURCE_PROPERTY_FORMAT TRUE)
    endif()

    get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)

    string(REPLACE ".cpp" "" EXE_NAME "${EXE_PREFIX}${SRC_FILENAME}")
    mfem_add_executable(${EXE_NAME} ${SRC_FILE})
-    install(TARGETS ${EXE_NAME}
-            RUNTIME DESTINATION examples)
    add_dependencies(${MFEM_ALL_EXAMPLES_TARGET_NAME} ${EXE_NAME})
    if (EXE_NEEDED_BY)
      add_dependencies(${EXE_NEEDED_BY} ${EXE_NAME})
@@ -156,6 +157,8 @@ macro(add_mfem_miniapp MFEM_EXE_NAME)
      endforeach()
      set(EXTRA_OPTIONS_LIST ${LIST_})
    endif()
+  elseif(MFEM_USE_HIP)
+    set_source_files_properties(${MAIN_LIST} ${EXTRA_SOURCES_LIST} PROPERTIES HIP_SOURCE_PROPERTY_FORMAT TRUE)
  endif()

  # Actually add the executable
@@ -529,15 +532,12 @@ function(mfem_find_package Name Prefix DirVar IncSuffixes Header LibSuffixes
            if (NOT ImportConfig)
              set(ImportConfig RELEASE)
            endif()
-            set(ImportConfigSuffix "_${ImportConfig}")
            get_target_property(ImpConfigs ${TargetName} IMPORTED_CONFIGURATIONS)
            list(FIND ImpConfigs ${ImportConfig} _Index)
-            if ((_Index EQUAL -1) OR ("${ImportConfig}" STREQUAL "NO_CONFIG"))
-              set(ImportConfig "NO_CONFIG")
-              set(ImportConfigSuffix "")
-              # message(FATAL_ERROR " *** ${ReqPack}: configuration "
-              #   "${ImportConfig} not found. Set ${ReqPack}_IMPORT_CONFIG "
-              #   "from the list: ${ImpConfigs}.")
+            if (_Index EQUAL -1)
+              message(FATAL_ERROR " *** ${ReqPack}: configuration "
+                "${ImportConfig} not found. Set ${ReqPack}_IMPORT_CONFIG "
+                "from the list: ${ImpConfigs}.")
            endif()
          endif()
          # Set _Pack_LIBS
@@ -549,8 +549,8 @@ function(mfem_find_package Name Prefix DirVar IncSuffixes Header LibSuffixes
            endif()
          else()
            # Set _Pack_LIBS from the target properties for ImportConfig
-            foreach (_prop IMPORTED_LOCATION${ImportConfigSuffix}
-                IMPORTED_LINK_INTERFACE_LIBRARIES${ImportConfigSuffix})
+            foreach (_prop IMPORTED_LOCATION_${ImportConfig}
+                IMPORTED_LINK_INTERFACE_LIBRARIES_${ImportConfig})
              get_target_property(_value ${TargetName} ${_prop})
              if (_value)
                list(APPEND _Pack_LIBS ${_value})
@@ -562,7 +562,7 @@ function(mfem_find_package Name Prefix DirVar IncSuffixes Header LibSuffixes
            endif()
          endif()
          # Set _Pack_INCS
-          foreach (_prop INCLUDE_DIRECTORIES INTERFACE_INCLUDE_DIRECTORIES)
+          foreach (_prop INCLUDE_DIRECTORIES)
            get_target_property(_value ${TargetName} ${_prop})
            if (_value)
              list(APPEND _Pack_INCS ${_value})
@@ -740,133 +740,6 @@ function(mfem_find_library Name Prefix Lib LibDoc CheckVar CheckSrc)
 endfunction(mfem_find_library)


-#
-# Extract compile and link options needed by the given target.
-#
-function(mfem_get_target_options Target CompileOptsVar LinkOptsVar)
-
-  if (NOT TARGET ${Target})
-    return()
-  endif()
-
-  # CMAKE_SHARED_LIBRARY_RUNTIME_C_FLAG -> '-Wl,-rpath,'
-  set(shared_link_flag ${CMAKE_SHARED_LIBRARY_RUNTIME_C_FLAG})
-  if (NOT shared_link_flag)
-    set(shared_link_flag "-Wl,-rpath,")
-  endif()
-
-  set(tgt "${Target}")
-  unset(CompileOpts)
-  unset(LinkOpts)
-  get_target_property(IsImported ${tgt} IMPORTED)
-  # message(STATUS "${tgt}[IMPORTED]: ${IsImported}")
-  # Generally, the possible target types are: STATIC_LIBRARY, MODULE_LIBRARY,
-  # SHARED_LIBRARY, INTERFACE_LIBRARY, EXECUTABLE.
-  get_target_property(type ${tgt} TYPE)
-  # message(STATUS "${tgt}[TYPE]: ${type}")
-  unset(ImportConfig)
-  get_target_property(ImportConfigs ${tgt} IMPORTED_CONFIGURATIONS)
-  if (ImportConfigs)
-    list(GET ImportConfigs 0 ImportConfig)
-  endif()
-  if (NOT ImportConfig)
-    set(ImportConfig RELEASE)
-  endif()
-  # message(STATUS "${tgt}[ImportConfig]: ${ImportConfig}")
-  # List all properties with: cmake --help-property-list
-  get_target_property(Defs ${tgt} INTERFACE_COMPILE_DEFINITIONS)
-  if (Defs)
-    list(REMOVE_DUPLICATES Defs)
-    foreach(Def ${Defs})
-      list(APPEND CompileOpts "-D${Def}")
-    endforeach()
-  endif()
-  get_target_property(Opts ${tgt} INTERFACE_COMPILE_OPTIONS)
-  if (Opts)
-    foreach(Opt ${Opts})
-      list(APPEND CompileOpts "${Opt}")
-    endforeach()
-  endif()
-  get_target_property(Dirs ${tgt} INTERFACE_INCLUDE_DIRECTORIES)
-  if (Dirs)
-    list(REMOVE_DUPLICATES Dirs)
-    foreach(Dir ${Dirs})
-      list(APPEND CompileOpts "-I\"${Dir}\"")
-    endforeach()
-  endif()
-  get_target_property(SysDirs ${tgt} INTERFACE_SYSTEM_INCLUDE_DIRECTORIES)
-  if (SysDirs)
-    list(REMOVE_DUPLICATES SysDirs)
-    foreach(SysDir ${SysDirs})
-      list(APPEND CompileOpts "-isystem \"${SysDir}\"")
-    endforeach()
-  endif()
-  if ("${type}" STREQUAL "STATIC_LIBRARY")
-    get_target_property(Location ${tgt} LOCATION)
-    if (Location)
-      list(APPEND LinkOpts "\"${Location}\"")
-    else()
-      message(STATUS " *** Warning: [${tgt}] LOCATION not defined!")
-    endif()
-  elseif ("${type}" STREQUAL "SHARED_LIBRARY")
-    get_target_property(Location ${tgt} LOCATION)
-    if (Location)
-      get_filename_component(Dir ${Location} DIRECTORY)
-      get_filename_component(NameWE ${Location} NAME_WE)
-      string(REGEX REPLACE "^lib" "" LibName ${NameWE})
-      list(APPEND LinkOpts
-        "-L\"${Dir}\""
-        "${shared_link_flag}\"${Dir}\""
-        "-l${LibName}")
-    else()
-      message(STATUS " *** Warning: [${tgt}] LOCATION not defined!")
-    endif()
-  elseif ("${type}" STREQUAL "INTERFACE_LIBRARY")
-    get_target_property(Libs ${tgt} INTERFACE_LINK_LIBRARIES)
-    if (Libs)
-      foreach(Lib ${Libs})
-        if (NOT (TARGET ${Lib}))
-          list(APPEND LinkOpts "${Lib}")
-        else()
-          mfem_get_target_options(${Lib} COpts LOpts)
-          list(APPEND CompileOpts ${COpts})
-          list(APPEND LinkOpts ${LOpts})
-        endif()
-      endforeach()
-    endif()
-    # Other properties we may need to handle:
-    # INTERFACE_LINK_DEPENDS
-    # INTERFACE_LINK_DIRECTORIES
-    # INTERFACE_LINK_OPTIONS
-  else()
-    message(STATUS " *** Warning: [${tgt}] uses target type '${type}'"
-      " which is not supported!")
-  endif()
-
-  # Other potentially relevant properties:
-  # - For all target types:
-  # IMPORTED_LIBNAME
-  # IMPORTED_LIBNAME_${ImportConfig}
-  # INTERFACE_AUTOUIC_OPTIONS
-  # INTERFACE_COMPILE_FEATURES
-  # INTERFACE_POSITION_INDEPENDENT_CODE
-  # INTERFACE_SOURCES
-  # INTERFACE_SYSTEM_INCLUDE_DIRECTORIES)
-  # - For non-"INTERFACE_LIBRARY" target types only:
-  # IMPORTED_LOCATION
-  # IMPORTED_LOCATION_${ImportConfig}
-  # IMPORTED_LINK_INTERFACE_LIBRARIES
-  # IMPORTED_LINK_INTERFACE_LIBRARIES_${ImportConfig}
-  # LINK_FLAGS
-  # LINK_FLAGS_${ImportConfig}
-  # LOCATION_${ImportConfig})
-
-  set(${CompileOptsVar} "${CompileOpts}" PARENT_SCOPE)
-  set(${LinkOptsVar} "${LinkOpts}" PARENT_SCOPE)
-
-endfunction(mfem_get_target_options)
-
-
 #
 #   Function that creates 'config.mk' from 'config.mk.in' for the both the
 #   build- and the install-locations and define install rules for 'config.mk'
@@ -885,15 +758,13 @@ function(mfem_export_mk_files)
  # Convert Boolean vars to YES/NO without writing the values to cache
  set(CONFIG_MK_BOOL_VARS MFEM_USE_MPI MFEM_USE_METIS MFEM_USE_METIS_5
      MFEM_DEBUG MFEM_USE_EXCEPTIONS MFEM_USE_ZLIB MFEM_USE_LIBUNWIND
-      MFEM_USE_LAPACK MFEM_THREAD_SAFE MFEM_USE_LEGACY_OPENMP MFEM_USE_OPENMP
+      MFEM_USE_LAPACK MFEM_THREAD_SAFE MFEM_USE_OPENMP MFEM_USE_LEGACY_OPENMP
      MFEM_USE_MEMALLOC MFEM_USE_SUNDIALS MFEM_USE_MESQUITE MFEM_USE_SUITESPARSE
-      MFEM_USE_SUPERLU MFEM_USE_SUPERLU5 MFEM_USE_MUMPS MFEM_USE_STRUMPACK
-      MFEM_USE_GINKGO MFEM_USE_AMGX MFEM_USE_GNUTLS MFEM_USE_NETCDF
-      MFEM_USE_PETSC MFEM_USE_SLEPC MFEM_USE_MPFR MFEM_USE_SIDRE MFEM_USE_FMS
-      MFEM_USE_CONDUIT MFEM_USE_PUMI MFEM_USE_HIOP MFEM_USE_GSLIB MFEM_USE_CUDA
-      MFEM_USE_HIP MFEM_USE_RAJA MFEM_USE_OCCA MFEM_USE_CEED MFEM_USE_CALIPER
-      MFEM_USE_UMPIRE MFEM_USE_SIMD MFEM_USE_ADIOS2 MFEM_USE_MKL_CPARDISO
-      MFEM_USE_ADFORWARD MFEM_USE_CODIPACK MFEM_USE_BENCHMARK MFEM_USE_PARELAG)
+      MFEM_USE_SUPERLU MFEM_USE_STRUMPACK MFEM_USE_GINKGO MFEM_USE_AMGX
+      MFEM_USE_GNUTLS MFEM_USE_GSLIB MFEM_USE_NETCDF MFEM_USE_PETSC
+      MFEM_USE_SLEPC MFEM_USE_MPFR MFEM_USE_SIDRE MFEM_USE_CONDUIT MFEM_USE_PUMI
+      MFEM_USE_CUDA MFEM_USE_OCCA MFEM_USE_RAJA MFEM_USE_UMPIRE MFEM_USE_SIMD
+      MFEM_USE_ADIOS2)
  foreach(var ${CONFIG_MK_BOOL_VARS})
    if (${var})
      set(${var} YES)
@@ -995,18 +866,8 @@ function(mfem_export_mk_files)
    get_filename_component(suffix ${lib} EXT)
    # handle interfaces (e.g., SCOREC::apf)
    if ("${lib}" MATCHES "SCOREC::.*" OR "${lib}" MATCHES "Ginkgo::.*")
-    elseif (TARGET "${lib}")
-      mfem_get_target_options(${lib} CompileOpts LinkOpts)
-      # Removing duplicates may lead to issues:
-      # list(REMOVE_DUPLICATES CompileOpts)
-      # list(REMOVE_DUPLICATES LinkOpts)
-      string(REPLACE ";" " " COpts "${CompileOpts}")
-      string(REPLACE ";" " " LOpts "${LinkOpts}")
-      # message(STATUS "${lib}[COpts]: '${COpts}'")
-      # message(STATUS "${lib}[LOpts]: '${LOpts}'")
-      set(MFEM_TPLFLAGS "${MFEM_TPLFLAGS} ${COpts}")
-      set(MFEM_EXT_LIBS "${MFEM_EXT_LIBS} ${LOpts}")
-      # message(FATAL_ERROR "***** interface lib found ... exiting *****")
+    elseif (NOT "${lib}" MATCHES "SCOREC::.*" AND "${lib}" MATCHES ".*::.*")
+      message(FATAL_ERROR "***** interface lib found ... exiting *****")
      # handle static and shared libs
    elseif ("${suffix}" STREQUAL "${CMAKE_SHARED_LIBRARY_SUFFIX}")
      get_filename_component(dir ${lib} DIRECTORY)
@@ -180,13 +180,4 @@
 // Enable interface to the MKL CPardiso library.
 // #define MFEM_USE_MKL_CPARDISO

-// Use forward mode for automatic differentiation
-// #define MFEM_USE_ADFORWARD
-
-// Enable the use of the CoDiPack library for AD
-// #define MFEM_USE_CODIPACK
-
-// Enable functionality based on the Google Benchmark library.
-// #define MFEM_USE_BENCHMARK
-
 #endif // MFEM_CONFIG_HEADER
@@ -58,10 +58,6 @@ MFEM_USE_UMPIRE        = @MFEM_USE_UMPIRE@
 MFEM_USE_SIMD          = @MFEM_USE_SIMD@
 MFEM_USE_ADIOS2        = @MFEM_USE_ADIOS2@
 MFEM_USE_MKL_CPARDISO  = @MFEM_USE_MKL_CPARDISO@
-MFEM_USE_ADFORWARD     = @MFEM_USE_ADFORWARD@
-MFEM_USE_CODIPACK      = @MFEM_USE_CODIPACK@
-MFEM_USE_BENCHMARK     = @MFEM_USE_BENCHMARK@
-MFEM_USE_PARELAG       = @MFEM_USE_PARELAG@

 # Compiler, compile options, and link options
 MFEM_CXX       = @MFEM_CXX@
@@ -91,7 +87,7 @@ MFEM_MPIEXEC_NP = @MFEM_MPIEXEC_NP@
 MFEM_MPI_NP     = @MFEM_MPI_NP@

 # The NVCC compiler cannot link with -x=cu
-MFEM_LINK_FLAGS := $(filter-out -x=cu -xhip, $(MFEM_FLAGS))
+MFEM_LINK_FLAGS := $(filter-out -x=cu, $(MFEM_FLAGS))

 # Optional extra configuration
@MFEM_CONFIG_EXTRA@
@@ -50,7 +50,6 @@ option(MFEM_USE_CONDUIT "Enable Conduit usage" OFF)
 option(MFEM_USE_PUMI "Enable PUMI" OFF)
 option(MFEM_USE_HIOP "Enable HiOp" OFF)
 option(MFEM_USE_CUDA "Enable CUDA" OFF)
-option(MFEM_USE_HIP "Enable HIP" OFF)
 option(MFEM_USE_OCCA "Enable OCCA" OFF)
 option(MFEM_USE_RAJA "Enable RAJA" OFF)
 option(MFEM_USE_CEED "Enable CEED" OFF)
@@ -59,10 +58,6 @@ option(MFEM_USE_SIMD "Enable use of SIMD intrinsics" OFF)
 option(MFEM_USE_ADIOS2 "Enable ADIOS2" OFF)
 option(MFEM_USE_CALIPER "Enable Caliper support" OFF)
 option(MFEM_USE_MKL_CPARDISO "Enable MKL CPardiso" OFF)
-option(MFEM_USE_ADFORWARD "Enable forward mode for AD" OFF)
-option(MFEM_USE_CODIPACK "Enable automatic differentiation (AD) using CoDiPack" OFF)
-option(MFEM_USE_BENCHMARK "Enable Google Benchmark" OFF)
-option(MFEM_USE_PARELAG "Enable ParELAG" OFF)

 # Optional overrides for autodetected MPIEXEC and MPIEXEC_NUMPROC_FLAG
 # set(MFEM_MPIEXEC "mpirun" CACHE STRING "Command for running MPI tests")
@@ -79,7 +74,6 @@ set(MFEM_MPI_NP 4 CACHE STRING "Number of processes used for MPI tests")
 option(MFEM_ENABLE_TESTING "Enable the ctest framework for testing" ON)
 option(MFEM_ENABLE_EXAMPLES "Build all of the examples" OFF)
 option(MFEM_ENABLE_MINIAPPS "Build all of the miniapps" OFF)
-option(MFEM_ENABLE_GOOGLE_BENCHMARKS "Build all of the Google benchmarks" OFF)

 # Setting CXX/MPICXX on the command line or in user.cmake will overwrite the
 # autodetected C++ compiler.
@@ -108,7 +102,6 @@ if (MFEM_USE_CUDA)
   set(HYPRE_REQUIRED_LIBRARIES "-lcusparse" "-lcurand" CACHE STRING
       "Libraries that HYPRE depends on.")
 endif()
-# HIP dependency for HYPRE is handled in FindHYPRE.cmake.

 set(METIS_DIR "${MFEM_DIR}/../metis-4.0" CACHE PATH "Path to the METIS library.")

@@ -228,34 +221,15 @@ set(MKL_LIBRARY_DIR "" CACHE STRING "Custom library subdirectory")

 set(OCCA_DIR "${MFEM_DIR}/../occa" CACHE PATH "Path to OCCA")
 set(RAJA_DIR "${MFEM_DIR}/../raja" CACHE PATH "Path to RAJA")
-# If RAJA is built with external CAMP:
-# set(RAJA_REQUIRED_PACKAGES "camp"
-#     CACHE STRING "Packages that RAJA depends on.")
-# set(camp_DIR "${MFEM_DIR}/../camp/lib/cmake/camp"
-#     CACHE PATH "Path to CAMP CMake files.")
 set(CEED_DIR "${MFEM_DIR}/../libCEED" CACHE PATH "Path to libCEED")
 set(UMPIRE_DIR "${MFEM_DIR}/../umpire" CACHE PATH "Path to Umpire")
 set(CALIPER_DIR "${MFEM_DIR}/../caliper" CACHE PATH "Path to Caliper")
-set(BENCHMARK_DIR "${MFEM_DIR}/../google-benchmark" CACHE PATH
-    "Path to Google Benchmark")
-
-# Provide paths, since ParELAG is dependent on MFEM and MFEM needs to be
-# compiled (or at least cmake needs to succeed) before compiling ParELAG.
-set(PARELAG_DIR "${MFEM_DIR}/../parelag" CACHE PATH "Path to ParELAG")
-set(PARELAG_INCLUDE_DIRS "${PARELAG_DIR}/src;${PARELAG_DIR}/build/src" CACHE
-    STRING "Path to ParELAG headers.")
-set(PARELAG_LIBRARIES "${PARELAG_DIR}/build/src/libParELAG.a" CACHE STRING
-    "The ParELAG library.")

 set(BLAS_INCLUDE_DIRS "" CACHE STRING "Path to BLAS headers.")
 set(BLAS_LIBRARIES "" CACHE STRING "The BLAS library.")
 set(LAPACK_INCLUDE_DIRS "" CACHE STRING "Path to LAPACK headers.")
 set(LAPACK_LIBRARIES "" CACHE STRING "The LAPACK library.")

-set(CODIPACK_INCLUDE_DIRS "${MFEM_DIR}/../CoDiPack/include" CACHE STRING
-    "Path to CoDiPack headers.")
-set(CODIPACK_LIBRARIES "")
-
 # Some useful variables:
 set(CMAKE_SKIP_PREPROCESSED_SOURCE_RULES ON) # Skip *.i rules
 set(CMAKE_SKIP_ASSEMBLY_SOURCE_RULES  ON)    # Skip *.s rules
@@ -59,12 +59,9 @@ HIP_FLAGS = --amdgpu-target=$(HIP_ARCH)
 HIP_XCOMPILER =
 HIP_XLINKER   = -Wl,

-# Flags for generating dependencies.
-DEP_FLAGS = -MM -MT
-
 ifneq ($(NOTMAC),)
   AR      = ar
-   ARFLAGS = crv
+   ARFLAGS = cruv
   RANLIB  = ranlib
   PICFLAG = $(XCOMPILER)-fPIC
   SO_EXT  = so
@@ -76,7 +73,7 @@ ifneq ($(NOTMAC),)
 else
   # Silence "has no symbols" warnings on Mac OS X
   AR      = ar
-   ARFLAGS = Scrv
+   ARFLAGS = Scruv
   RANLIB  = ranlib -no_warning_for_no_symbols
   PICFLAG = $(XCOMPILER)-fPIC
   SO_EXT  = dylib
@@ -89,9 +86,6 @@ else
   BUILD_RPATH = $(XLINKER)-undefined,dynamic_lookup
   INSTALL_SOFLAGS = $(subst $1 ,,$(call MAKE_SOFLAGS,$(MFEM_LIB_DIR)))
   INSTALL_RPATH = $(XLINKER)-undefined,dynamic_lookup
-   # Silence unused command line argument warnings when generating dependencies
-   # with mpicxx and clang
-   DEP_FLAGS := -Wno-unused-command-line-argument $(DEP_FLAGS)
 endif

 # Set CXXFLAGS to overwrite the default selection of DEBUG_FLAGS/OPTIM_FLAGS
@@ -157,10 +151,6 @@ MFEM_USE_UMPIRE        = NO
 MFEM_USE_SIMD          = NO
 MFEM_USE_ADIOS2        = NO
 MFEM_USE_MKL_CPARDISO  = NO
-MFEM_USE_ADFORWARD     = NO
-MFEM_USE_CODIPACK      = NO
-MFEM_USE_BENCHMARK     = NO
-MFEM_USE_PARELAG       = NO

 # MPI library compile and link flags
 # These settings are used only when building MFEM with MPI + HIP
@@ -172,20 +162,6 @@ ifeq ($(MFEM_USE_MPI)$(MFEM_USE_HIP),YESYES)
   MPI_LIB = -L$(MPI_DIR)/lib $(XLINKER)-rpath,$(MPI_DIR)/lib -lmpi
 endif

-# ROCM/HIP directory such that ROCM/HIP libraries like rocsparse and rocrand are
-# found in $(HIP_DIR)/lib, usually as links. Typically, this directoory is of
-# the form /opt/rocm-X.Y.Z which is called ROCM_PATH by hipconfig.
-ifeq ($(MFEM_USE_HIP),YES)
-   HIP_DIR := $(patsubst %/,%,$(dir $(shell which $(HIP_CXX))))
-   HIP_DIR := $(patsubst %/,%,$(dir $(HIP_DIR)))
-   ifeq (,$(wildcard $(HIP_DIR)/lib/librocsparse.*))
-      HIP_DIR := $(shell hipconfig --rocmpath 2> /dev/null)
-      ifeq (,$(wildcard $(HIP_DIR)/lib/librocsparse.*))
-         $(error Unable to determine HIP_DIR. Please set it manually.)
-      endif
-   endif
-endif
-
 # Compile and link options for zlib.
 ZLIB_DIR =
 ZLIB_OPT = $(if $(ZLIB_DIR),-I$(ZLIB_DIR)/include)
@@ -203,11 +179,6 @@ ifeq (YES,$(MFEM_USE_CUDA))
   # This is only necessary when hypre is built with cuda:
   HYPRE_LIB += -lcusparse -lcurand
 endif
-ifeq (YES,$(MFEM_USE_HIP))
-   # This is only necessary when hypre is built with hip:
-   HYPRE_LIB += -L$(HIP_DIR)/lib $(XLINKER)-rpath,$(HIP_DIR)/lib\
- -lrocsparse -lrocrand
-endif

 # METIS library configuration
 ifeq ($(MFEM_USE_SUPERLU)$(MFEM_USE_STRUMPACK)$(MFEM_USE_MUMPS),NONONO)
@@ -435,11 +406,6 @@ HIOP_DIR = @MFEM_DIR@/../hiop/install
 HIOP_OPT = -I$(HIOP_DIR)/include
 HIOP_LIB = -L$(HIOP_DIR)/lib -lhiop $(LAPACK_LIB)

-# CoDiPack
-CODIPACK_DIR = @MFEM_DIR@/../CoDiPack
-CODIPACK_OPT = -I$(CODIPACK_DIR)
-CODIPACK_LIB =
-
 # GSLIB library
 GSLIB_DIR = @MFEM_DIR@/../gslib/build
 GSLIB_OPT = -I$(GSLIB_DIR)/include
@@ -449,9 +415,9 @@ GSLIB_LIB = -L$(GSLIB_DIR)/lib -lgs
 CUDA_OPT =
 CUDA_LIB = -lcusparse

-# HIP library configuration
+# HIP library configuration (currently not needed)
 HIP_OPT =
-HIP_LIB = -L$(HIP_DIR)/lib $(XLINKER)-rpath,$(HIP_DIR)/lib -lhipsparse
+HIP_LIB =

 # OCCA library configuration
 OCCA_DIR = @MFEM_DIR@/../occa
@@ -463,11 +429,6 @@ CALIPER_DIR = @MFEM_DIR@/../caliper
 CALIPER_OPT = -I$(CALIPER_DIR)/include
 CALIPER_LIB = $(XLINKER)-rpath,$(CALIPER_DIR)/lib64 -L$(CALIPER_DIR)/lib64 -lcaliper

-# BENCHMARK library configuration
-BENCHMARK_DIR = @MFEM_DIR@/../google-benchmark
-BENCHMARK_OPT = -I$(BENCHMARK_DIR)/include
-BENCHMARK_LIB = -L$(BENCHMARK_DIR)/lib -lbenchmark -lpthread
-
 # libCEED library configuration
 CEED_DIR ?= @MFEM_DIR@/../libCEED
 CEED_OPT = -I$(CEED_DIR)/include
@@ -498,11 +459,6 @@ MKL_CPARDISO_LIB = $(XLINKER)-rpath,$(MKL_CPARDISO_DIR)/$(MKL_LIBRARY_SUBDIR)\
   -L$(MKL_CPARDISO_DIR)/$(MKL_LIBRARY_SUBDIR) -l$(MKL_MPI_WRAPPER)\
   -lmkl_intel_lp64 -lmkl_sequential -lmkl_core

-# PARELAG library configuration
-PARELAG_DIR = @MFEM_DIR@/../parelag
-PARELAG_OPT = -I$(PARELAG_DIR)/src -I$(PARELAG_DIR)/build/src
-PARELAG_LIB = -L$(PARELAG_DIR)/build/src -lParELAG
-
 # If YES, enable some informational messages
 VERBOSE = NO

@@ -87,12 +87,12 @@ fi
 ## style check
 #if [[ "${option}" == "--style" || "${option}" == "" ]]; then
 if [[ "${option}" == "--style" ]]; then
-  if which astyle && [[ "$(astyle --version)" == "Artistic Style Version 3.1" ]]; then
+  if which astyle && [[ "$(astyle --version)" == "Artistic Style Version 2.05.1" ]]; then
    cd tests/scripts
    if ! ./runtest code-style; then code=1; fi
    cd -
  else
-    echo "Warning: astyle not found or version is not 3.1"
+    echo "Warning: astyle not found or version is not 2.05.1"
  fi
 fi

@@ -58,14 +58,14 @@ ifneq (,$(filter test%,$(MAKECMDGOALS)))
   MAKEFLAGS += -k
 endif
 # Test runs of the examples/miniapps with parameters - check exit code:
-# 0 means success, 242 means the test was skipped, anything else means error
+# 0 means success, 255 means the test was skipped, anything else means error
 mfem-test = \
   printf "   $(3) [$(2) $(1) ... ]: "; \
   $(call $(TIMEFUN),$(TIMECMD),$(2) ./$(1) $(if $(5),,-no-vis )$(4) \
     > $(1).stderr 2>&1); \
   err="$$3"; \
   if [ "$$3" = 0 ]; then $(PRINT_OK); \
-   else if [ "$$3" = 242 ]; then $(PRINT_SKIP); err=0; \
+   else if [ "$$3" = 255 ]; then $(PRINT_SKIP); err=0; \
   else $(PRINT_FAILED); cat $(1).stderr; fi; fi; \
   rm -f $(1).stderr; exit $$err

@@ -76,7 +76,7 @@ mfem-test-file = \
   $(call $(TIMEFUN),$(TIMECMD),$(2) ./$(1) -no-vis > $(1).stderr 2>&1); \
   err="$$3"; \
   if [ "$$3" = 0 ] && [ -e $(4) ]; then $(PRINT_OK); \
-   else if [ "$$3" = 242 ] && [ -e $(4) ]; then $(PRINT_SKIP); err=0; \
+   else if [ "$$3" = 255 ] && [ -e $(4) ]; then $(PRINT_SKIP); err=0; \
   else $(PRINT_FAILED); cat $(1).stderr; err=64; fi; fi; \
   rm -f $(1).stderr; exit $$err

@@ -765,7 +765,6 @@ INPUT                  = @MFEM_SOURCE_DIR@/doc/CodeDocumentation.dox \
                         @MFEM_SOURCE_DIR@/linalg \
                         @MFEM_SOURCE_DIR@/mesh \
                         @MFEM_SOURCE_DIR@/fem \
-                         @MFEM_SOURCE_DIR@/fem/fe \
                         @MFEM_SOURCE_DIR@/examples \
                         @MFEM_SOURCE_DIR@/examples/caliper \
                         @MFEM_SOURCE_DIR@/examples/amgx \
@@ -781,15 +780,13 @@ INPUT                  = @MFEM_SOURCE_DIR@/doc/CodeDocumentation.dox \
                         @MFEM_SOURCE_DIR@/miniapps/gslib \
                         @MFEM_SOURCE_DIR@/miniapps/meshing \
                         @MFEM_SOURCE_DIR@/miniapps/mtop \
-                         @MFEM_SOURCE_DIR@/miniapps/autodiff \
                         @MFEM_SOURCE_DIR@/miniapps/navier \
                         @MFEM_SOURCE_DIR@/miniapps/nurbs \
                         @MFEM_SOURCE_DIR@/miniapps/performance \
                         @MFEM_SOURCE_DIR@/miniapps/shifted \
                         @MFEM_SOURCE_DIR@/miniapps/solvers \
                         @MFEM_SOURCE_DIR@/miniapps/tools \
-                         @MFEM_SOURCE_DIR@/miniapps/toys \
-                         @MFEM_SOURCE_DIR@/miniapps/parelag
+                         @MFEM_SOURCE_DIR@/miniapps/toys

 # This tag can be used to specify the character encoding of the source files
 # that doxygen parses. Internally doxygen uses the UTF-8 encoding. Doxygen uses
@@ -194,8 +194,6 @@ namespace mfem {
 * - <a class="el" href="parheat_8cpp_source.html">Optimization gradients</a>: Gradients of PDE-constrained function
 * - <a class="el" href="miniapps_2performance_2ex1_8cpp_source.html">HPC Example 1</a>: high-performance nodal H1 FEM for the Laplace problem
 * - <a class="el" href="miniapps_2performance_2ex1p_8cpp_source.html">HPC Example 1p</a>: high-performance parallel nodal H1 FEM for the Laplace problem
- * - <a class="el" href="par__example_8cpp_source.html">Parallel pLaplacian example using AD</a>: Parallel pLaplacian example
- * - <a class="el" href="seq__example_8cpp_source.html">Serial pLaplacian example using AD</a>: Serial pLaplacian example
 *
 * See also the <a class="el" href="https://mfem.org/examples/">examples documentation</a> online.
 */
@@ -37,7 +37,6 @@ list(APPEND ALL_EXE_SRCS
  ex27.cpp
  ex28.cpp
  ex29.cpp
-  ex30.cpp
  )

 if (MFEM_USE_MPI)
@@ -71,7 +70,6 @@ if (MFEM_USE_MPI)
    ex27p.cpp
    ex28p.cpp
    ex29p.cpp
-    ex30p.cpp
    )
 endif()

@@ -82,80 +80,78 @@ include_directories(BEFORE ${PROJECT_BINARY_DIR})
 add_mfem_examples(ALL_EXE_SRCS)

 # Add a test for each example
-if (MFEM_ENABLE_TESTING)
-  foreach(SRC_FILE ${ALL_EXE_SRCS})
-    get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
-    string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
+foreach(SRC_FILE ${ALL_EXE_SRCS})
+  get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
+  string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})

-    set(THIS_TEST_OPTIONS "-no-vis")
-    if (${TEST_NAME} MATCHES "ex0p?")
-      set(THIS_TEST_OPTIONS)
-    endif()
-    if (${TEST_NAME} MATCHES "ex10p*")
-      list(APPEND THIS_TEST_OPTIONS "-tf" "5")
-    elseif(${TEST_NAME} MATCHES "ex15p*")
-      list(APPEND THIS_TEST_OPTIONS "-e" "1")
-    elseif(${TEST_NAME} MATCHES "ex27p*")
-      list(APPEND THIS_TEST_OPTIONS "-dg")
-    endif()
+  set(THIS_TEST_OPTIONS "-no-vis")
+  if (${TEST_NAME} MATCHES "ex0p?")
+    set(THIS_TEST_OPTIONS)
+  endif()
+  if (${TEST_NAME} MATCHES "ex10p*")
+    list(APPEND THIS_TEST_OPTIONS "-tf" "5")
+  elseif(${TEST_NAME} MATCHES "ex15p*")
+    list(APPEND THIS_TEST_OPTIONS "-e" "1")
+  elseif(${TEST_NAME} MATCHES "ex27p*")
+    list(APPEND THIS_TEST_OPTIONS "-dg")
+  endif()

+  if (NOT (${TEST_NAME} MATCHES ".*p$"))
+    add_test(NAME ${TEST_NAME}_ser
+      COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
+  else()
+    add_test(NAME ${TEST_NAME}_np=${MFEM_MPI_NP}
+      COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
+      ${MPIEXEC_PREFLAGS}
+      $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
+      ${MPIEXEC_POSTFLAGS})
+  endif()
+endforeach()
+
+# Add CUDA/HIP tests.
+set(DEVICE_EXAMPLES
+    # serial examples with device support:
+    ex1 ex3 ex4 ex5 ex6 ex9 ex22 ex24 ex25 ex26
+    # parallel examples with device support:
+    ex1p ex2p ex3p ex4p ex5p ex6p ex7p ex9p ex13p ex22p ex24p ex25p ex26p)
+set(MFEM_TEST_DEVICE)
+if (MFEM_USE_CUDA)
+  set(MFEM_TEST_DEVICE "cuda")
+elseif (MFEM_USE_HIP)
+  set(MFEM_TEST_DEVICE "hip")
+endif()
+if (MFEM_TEST_DEVICE)
+  foreach(TEST_NAME ${DEVICE_EXAMPLES})
+    set(THIS_TEST_OPTIONS "-no-vis" "-d" "${MFEM_TEST_DEVICE}")
    if (NOT (${TEST_NAME} MATCHES ".*p$"))
-      add_test(NAME ${TEST_NAME}_ser
+      add_test(NAME ${TEST_NAME}_${MFEM_TEST_DEVICE}_ser
        COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
-    else()
-      add_test(NAME ${TEST_NAME}_np=${MFEM_MPI_NP}
+    elseif (MFEM_USE_MPI)
+      add_test(NAME ${TEST_NAME}_${MFEM_TEST_DEVICE}_np=${MFEM_MPI_NP}
        COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
        ${MPIEXEC_PREFLAGS}
        $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
        ${MPIEXEC_POSTFLAGS})
    endif()
  endforeach()
+endif()

-  # Add CUDA/HIP tests.
-  set(DEVICE_EXAMPLES
-      # serial examples with device support:
-      ex1 ex3 ex4 ex5 ex6 ex9 ex22 ex24 ex25 ex26
-      # parallel examples with device support:
-      ex1p ex2p ex3p ex4p ex5p ex6p ex7p ex9p ex13p ex22p ex24p ex25p ex26p)
-  set(MFEM_TEST_DEVICE)
-  if (MFEM_USE_CUDA)
-    set(MFEM_TEST_DEVICE "cuda")
-  elseif (MFEM_USE_HIP)
-    set(MFEM_TEST_DEVICE "hip")
-  endif()
-  if (MFEM_TEST_DEVICE)
-    foreach(TEST_NAME ${DEVICE_EXAMPLES})
-      set(THIS_TEST_OPTIONS "-no-vis" "-d" "${MFEM_TEST_DEVICE}")
-      if (NOT (${TEST_NAME} MATCHES ".*p$"))
-        add_test(NAME ${TEST_NAME}_${MFEM_TEST_DEVICE}_ser
-          COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
-      elseif (MFEM_USE_MPI)
-        add_test(NAME ${TEST_NAME}_${MFEM_TEST_DEVICE}_np=${MFEM_MPI_NP}
-          COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
-          ${MPIEXEC_PREFLAGS}
-          $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
-          ${MPIEXEC_POSTFLAGS})
-      endif()
-    endforeach()
-  endif()
+# If STRUMPACK is enabled, add a test run that uses it.
+if (MFEM_USE_STRUMPACK)
+  add_test(NAME ex11p_strumpack_np=${MFEM_MPI_NP}
+    COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
+    ${MPIEXEC_PREFLAGS}
+    $<TARGET_FILE:ex11p> "-no-vis" "--strumpack"
+    ${MPIEXEC_POSTFLAGS})
+endif()

-  # If STRUMPACK is enabled, add a test run that uses it.
-  if (MFEM_USE_STRUMPACK)
-    add_test(NAME ex11p_strumpack_np=${MFEM_MPI_NP}
-      COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
-      ${MPIEXEC_PREFLAGS}
-      $<TARGET_FILE:ex11p> "-no-vis" "--strumpack"
-      ${MPIEXEC_POSTFLAGS})
-  endif()
-
-  # If SuperLU_DIST is enabled, add a test run that uses it.
-  if (MFEM_USE_SUPERLU)
-    add_test(NAME ex11p_superlu_np=${MFEM_MPI_NP}
-      COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
-      ${MPIEXEC_PREFLAGS}
-      $<TARGET_FILE:ex11p> "-no-vis" "--superlu"
-      ${MPIEXEC_POSTFLAGS})
-  endif()
+# If SuperLU_DIST is enabled, add a test run that uses it.
+if (MFEM_USE_SUPERLU)
+  add_test(NAME ex11p_superlu_np=${MFEM_MPI_NP}
+    COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
+    ${MPIEXEC_PREFLAGS}
+    $<TARGET_FILE:ex11p> "-no-vis" "--superlu"
+    ${MPIEXEC_POSTFLAGS})
 endif()

 # Include the examples/amgx directory if AmgX is enabled
@@ -50,32 +50,30 @@ add_mfem_examples(AMGX_EXAMPLES_SRCS ${PFX} copy_amgx_json_files test_amgx)
 # which builds the examples and runs:
 #   ctest -R amgx

-if (MFEM_ENABLE_TESTING)
-  # Command line options for the tests.
-  # Example 1/1p:
-  set(EX1_TEST_OPTS)
-  set(EX1P_TEST_OPTS)
+# Command line options for the tests.
+# Example 1/1p:
+set(EX1_TEST_OPTS)
+set(EX1P_TEST_OPTS)

-  # Add the tests: one test per source file.
-  foreach(SRC_FILE ${AMGX_EXAMPLES_SRCS})
-    get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
-    string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
-    string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
-    set(TEST_NAME ${PFX}${TEST_NAME})
+# Add the tests: one test per source file.
+foreach(SRC_FILE ${AMGX_EXAMPLES_SRCS})
+  get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
+  string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
+  string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
+  set(TEST_NAME ${PFX}${TEST_NAME})

-    set(THIS_TEST_OPTIONS "-no-vis")
-    list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
-    # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")
+  set(THIS_TEST_OPTIONS "-no-vis")
+  list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
+  # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")

-    if (NOT (${TEST_NAME} MATCHES ".*p$"))
-      add_test(NAME ${TEST_NAME}_ser
-        COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
-    else()
-      add_test(NAME ${TEST_NAME}_np=${MFEM_MPI_NP}
-        COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
-        ${MPIEXEC_PREFLAGS}
-        $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
-        ${MPIEXEC_POSTFLAGS})
-    endif()
-  endforeach()
-endif()
+  if (NOT (${TEST_NAME} MATCHES ".*p$"))
+    add_test(NAME ${TEST_NAME}_ser
+      COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
+  else()
+    add_test(NAME ${TEST_NAME}_np=${MFEM_MPI_NP}
+      COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
+      ${MPIEXEC_PREFLAGS}
+      $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
+      ${MPIEXEC_POSTFLAGS})
+  endif()
+endforeach()
@@ -30,22 +30,22 @@ set(PREFIX caliper_)
 add_mfem_examples(CALIPER_EXE_SRCS ${PREFIX})

 # Add a test for each example
-if (MFEM_ENABLE_TESTING)
-  foreach(SRC_FILE ${CALIPER_EXE_SRCS})
-    get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
-    string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
+foreach(SRC_FILE ${CALIPER_EXE_SRCS})
+  get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
+  string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
+
+  set(THIS_TEST_OPTIONS "-no-vis")
+
+  if (NOT (${TEST_NAME} MATCHES ".*p$"))
+    add_test(NAME ${TEST_NAME}_ser
+      COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
+  else()
+    add_test(NAME ${TEST_NAME}_np=${MFEM_MPI_NP}
+      COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
+      ${MPIEXEC_PREFLAGS}
+      $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
+      ${MPIEXEC_POSTFLAGS})
+  endif()
+endforeach()

-    set(THIS_TEST_OPTIONS "-no-vis")

-    if (NOT (${TEST_NAME} MATCHES ".*p$"))
-      add_test(NAME ${TEST_NAME}_ser
-        COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
-    else()
-      add_test(NAME ${TEST_NAME}_np=${MFEM_MPI_NP}
-        COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
-        ${MPIEXEC_PREFLAGS}
-        $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
-        ${MPIEXEC_POSTFLAGS})
-    endif()
-  endforeach()
-endif()
@@ -149,7 +149,7 @@ void InitialDeformation(const Vector &x, Vector &y);

 void InitialVelocity(const Vector &x, Vector &v);

-void visualize(ostream &os, Mesh *mesh, GridFunction *deformed_nodes,
+void visualize(ostream &out, Mesh *mesh, GridFunction *deformed_nodes,
               GridFunction *field, const char *field_name = NULL,
               bool init_vis = false);

@@ -376,10 +376,10 @@ int main(int argc, char *argv[])
 }


-void visualize(ostream &os, Mesh *mesh, GridFunction *deformed_nodes,
+void visualize(ostream &out, Mesh *mesh, GridFunction *deformed_nodes,
               GridFunction *field, const char *field_name, bool init_vis)
 {
-   if (!os)
+   if (!out)
   {
      return;
   }
@@ -389,25 +389,24 @@ void visualize(ostream &os, Mesh *mesh, GridFunction *deformed_nodes,

   mesh->SwapNodes(nodes, owns_nodes);

-   os << "solution\n" << *mesh << *field;
+   out << "solution\n" << *mesh << *field;

   mesh->SwapNodes(nodes, owns_nodes);

   if (init_vis)
   {
-      os << "window_size 800 800\n";
-      os << "window_title '" << field_name << "'\n";
+      out << "window_size 800 800\n";
+      out << "window_title '" << field_name << "'\n";
      if (mesh->SpaceDimension() == 2)
      {
-         os << "view 0 0\n"; // view from top
-         os << "keys jl\n";  // turn off perspective and light
+         out << "view 0 0\n"; // view from top
+         out << "keys jl\n";  // turn off perspective and light
      }
-      os << "keys cm\n";         // show colorbar and mesh
-      // update value-range; keep mesh-extents fixed
-      os << "autoscale value\n";
-      os << "pause\n";
+      out << "keys cm\n";         // show colorbar and mesh
+      out << "autoscale value\n"; // update value-range; keep mesh-extents fixed
+      out << "pause\n";
   }
-   os << flush;
+   out << flush;
 }


@@ -154,8 +154,7 @@ void InitialDeformation(const Vector &x, Vector &y);

 void InitialVelocity(const Vector &x, Vector &v);

-void visualize(ostream &os, ParMesh *mesh,
-               ParGridFunction *deformed_nodes,
+void visualize(ostream &out, ParMesh *mesh, ParGridFunction *deformed_nodes,
               ParGridFunction *field, const char *field_name = NULL,
               bool init_vis = false);

@@ -439,11 +438,10 @@ int main(int argc, char *argv[])
   return 0;
 }

-void visualize(ostream &os, ParMesh *mesh,
-               ParGridFunction *deformed_nodes,
+void visualize(ostream &out, ParMesh *mesh, ParGridFunction *deformed_nodes,
               ParGridFunction *field, const char *field_name, bool init_vis)
 {
-   if (!os)
+   if (!out)
   {
      return;
   }
@@ -453,27 +451,25 @@ void visualize(ostream &os, ParMesh *mesh,

   mesh->SwapNodes(nodes, owns_nodes);

-   os << "parallel " << mesh->GetNRanks()
-      << " " << mesh->GetMyRank() << "\n";
-   os << "solution\n" << *mesh << *field;
+   out << "parallel " << mesh->GetNRanks() << " " << mesh->GetMyRank() << "\n";
+   out << "solution\n" << *mesh << *field;

   mesh->SwapNodes(nodes, owns_nodes);

   if (init_vis)
   {
-      os << "window_size 800 800\n";
-      os << "window_title '" << field_name << "'\n";
+      out << "window_size 800 800\n";
+      out << "window_title '" << field_name << "'\n";
      if (mesh->SpaceDimension() == 2)
      {
-         os << "view 0 0\n"; // view from top
-         os << "keys jl\n";  // turn off perspective and light
+         out << "view 0 0\n"; // view from top
+         out << "keys jl\n";  // turn off perspective and light
      }
-      os << "keys cm\n";         // show colorbar and mesh
-      // update value-range; keep mesh-extents fixed
-      os << "autoscale value\n";
-      os << "pause\n";
+      out << "keys cm\n";         // show colorbar and mesh
+      out << "autoscale value\n"; // update value-range; keep mesh-extents fixed
+      out << "pause\n";
   }
-   os << flush;
+   out << flush;
 }


@@ -135,8 +135,8 @@ int main(int argc, char *argv[])
   a->AddBdrFaceIntegrator(new DGDiffusionIntegrator(one, sigma, kappa));
   if (eta > 0)
   {
-      a->AddInteriorFaceIntegrator(new DGDiffusionBR2Integrator(*fespace, eta));
-      a->AddBdrFaceIntegrator(new DGDiffusionBR2Integrator(*fespace, eta));
+      a->AddInteriorFaceIntegrator(new DGDiffusionBR2Integrator(fespace, eta));
+      a->AddBdrFaceIntegrator(new DGDiffusionBR2Integrator(fespace, eta));
   }
   a->Assemble();
   a->Finalize();
@@ -199,8 +199,8 @@ int main(int argc, char *argv[])
   a->AddBdrFaceIntegrator(new DGDiffusionIntegrator(one, sigma, kappa));
   if (eta > 0)
   {
-      a->AddInteriorFaceIntegrator(new DGDiffusionBR2Integrator(*fespace, eta));
-      a->AddBdrFaceIntegrator(new DGDiffusionBR2Integrator(*fespace, eta));
+      a->AddInteriorFaceIntegrator(new DGDiffusionBR2Integrator(fespace, eta));
+      a->AddBdrFaceIntegrator(new DGDiffusionBR2Integrator(fespace, eta));
   }
   a->Assemble();
   a->Finalize();
@@ -221,7 +221,7 @@ int main(int argc, char *argv[])
   {
      HyprePCG pcg(*A);
      pcg.SetTol(1e-12);
-      pcg.SetMaxIter(500);
+      pcg.SetMaxIter(200);
      pcg.SetPrintLevel(2);
      pcg.SetPreconditioner(*amg);
      pcg.Mult(*B, *X);
@@ -232,7 +232,7 @@ int main(int argc, char *argv[])
      GMRESSolver gmres(MPI_COMM_WORLD);
      gmres.SetAbsTol(0.0);
      gmres.SetRelTol(1e-12);
-      gmres.SetMaxIter(500);
+      gmres.SetMaxIter(200);
      gmres.SetKDim(10);
      gmres.SetPrintLevel(1);
      gmres.SetOperator(*A);
@@ -32,7 +32,7 @@ private:
   mutable DenseTensor flux;
   mutable Vector z;

-   void GetFlux(const DenseMatrix &state_, DenseTensor &flux_) const;
+   void GetFlux(const DenseMatrix &state, DenseTensor &flux) const;

 public:
   FE_Evolution(FiniteElementSpace &vfes_,
@@ -256,26 +256,26 @@ inline double ComputeMaxCharSpeed(const Vector &state, const int dim)
 }

 // Compute the flux at solution nodes.
-void FE_Evolution::GetFlux(const DenseMatrix &x_, DenseTensor &flux_) const
+void FE_Evolution::GetFlux(const DenseMatrix &x, DenseTensor &flux) const
 {
-   const int flux_dof = flux_.SizeI();
-   const int flux_dim = flux_.SizeJ();
+   const int dof = flux.SizeI();
+   const int dim = flux.SizeJ();

-   for (int i = 0; i < flux_dof; i++)
+   for (int i = 0; i < dof; i++)
   {
-      for (int k = 0; k < num_equation; k++) { state(k) = x_(i, k); }
-      ComputeFlux(state, flux_dim, f);
+      for (int k = 0; k < num_equation; k++) { state(k) = x(i, k); }
+      ComputeFlux(state, dim, f);

-      for (int d = 0; d < flux_dim; d++)
+      for (int d = 0; d < dim; d++)
      {
         for (int k = 0; k < num_equation; k++)
         {
-            flux_(i, d, k) = f(k, d);
+            flux(i, d, k) = f(k, d);
         }
      }

      // Update max char speed
-      const double mcs = ComputeMaxCharSpeed(state, flux_dim);
+      const double mcs = ComputeMaxCharSpeed(state, dim);
      if (mcs > max_char_speed) { max_char_speed = mcs; }
   }
 }
@@ -171,7 +171,7 @@ public:
 };

 // Visualization driver
-void visualize(ostream &os, Mesh *mesh, GridFunction *deformed_nodes,
+void visualize(ostream &out, Mesh *mesh, GridFunction *deformed_nodes,
               GridFunction *field, const char *field_name = NULL,
               bool init_vis = false);

@@ -542,10 +542,10 @@ RubberOperator::~RubberOperator()


 // Inline visualization
-void visualize(ostream &os, Mesh *mesh, GridFunction *deformed_nodes,
+void visualize(ostream &out, Mesh *mesh, GridFunction *deformed_nodes,
               GridFunction *field, const char *field_name, bool init_vis)
 {
-   if (!os)
+   if (!out)
   {
      return;
   }
@@ -555,25 +555,23 @@ void visualize(ostream &os, Mesh *mesh, GridFunction *deformed_nodes,

   mesh->SwapNodes(nodes, owns_nodes);

-   os << "solution\n" << *mesh << *field;
+   out << "solution\n" << *mesh << *field;

   mesh->SwapNodes(nodes, owns_nodes);

   if (init_vis)
   {
-      os << "window_size 800 800\n";
-      os << "window_title '" << field_name << "'\n";
+      out << "window_size 800 800\n";
+      out << "window_title '" << field_name << "'\n";
      if (mesh->SpaceDimension() == 2)
      {
-         os << "view 0 0\n"; // view from top
-         // turn off perspective and light, +anti-aliasing
-         os << "keys jlA\n";
+         out << "view 0 0\n"; // view from top
+         out << "keys jlA\n"; // turn off perspective and light, +anti-aliasing
      }
-      os << "keys cmA\n"; // show colorbar and mesh, +anti-aliasing
-      // update value-range; keep mesh-extents fixed
-      os << "autoscale value\n";
+      out << "keys cmA\n";        // show colorbar and mesh, +anti-aliasing
+      out << "autoscale value\n"; // update value-range; keep mesh-extents fixed
   }
-   os << flush;
+   out << flush;
 }

 void ReferenceConfiguration(const Vector &x, Vector &y)
@@ -185,8 +185,7 @@ public:
 };

 // Visualization driver
-void visualize(ostream &os, ParMesh *mesh,
-               ParGridFunction *deformed_nodes,
+void visualize(ostream &out, ParMesh *mesh, ParGridFunction *deformed_nodes,
               ParGridFunction *field, const char *field_name = NULL,
               bool init_vis = false);

@@ -197,10 +196,10 @@ void InitialDeformation(const Vector &x, Vector &y);

 int main(int argc, char *argv[])
 {
-#ifdef HYPRE_USING_GPU
+#ifdef HYPRE_USING_CUDA
   cout << "\nAs of mfem-4.3 and hypre-2.22.0 (July 2021) this example\n"
-        << "is NOT supported with the GPU version of hypre.\n\n";
-   return 242;
+        << "is NOT supported with the CUDA version of hypre.\n\n";
+   return 255;
 #endif

   // 1. Initialize MPI
@@ -487,8 +486,8 @@ void JacobianPreconditioner::SetOperator(const Operator &op)

      if (!spaces[0]->GetParMesh()->Nonconforming())
      {
-#if !defined(HYPRE_USING_GPU)
-         // Not available yet when hypre is built with GPU support
+#ifndef HYPRE_USING_CUDA
+         // Not available yet when hypre is built with CUDA
         stiff_prec_amg->SetElasticityOptions(spaces[0]);
 #endif
      }
@@ -618,11 +617,10 @@ RubberOperator::~RubberOperator()


 // Inline visualization
-void visualize(ostream &os, ParMesh *mesh,
-               ParGridFunction *deformed_nodes,
+void visualize(ostream &out, ParMesh *mesh, ParGridFunction *deformed_nodes,
               ParGridFunction *field, const char *field_name, bool init_vis)
 {
-   if (!os)
+   if (!out)
   {
      return;
   }
@@ -632,27 +630,24 @@ void visualize(ostream &os, ParMesh *mesh,

   mesh->SwapNodes(nodes, owns_nodes);

-   os << "parallel " << mesh->GetNRanks() << " " << mesh->GetMyRank() <<
-      "\n";
-   os << "solution\n" << *mesh << *field;
+   out << "parallel " << mesh->GetNRanks() << " " << mesh->GetMyRank() << "\n";
+   out << "solution\n" << *mesh << *field;

   mesh->SwapNodes(nodes, owns_nodes);

   if (init_vis)
   {
-      os << "window_size 800 800\n";
-      os << "window_title '" << field_name << "'\n";
+      out << "window_size 800 800\n";
+      out << "window_title '" << field_name << "'\n";
      if (mesh->SpaceDimension() == 2)
      {
-         os << "view 0 0\n"; // view from top
-         // turn off perspective and light, +anti-aliasing
-         os << "keys jlA\n";
+         out << "view 0 0\n"; // view from top
+         out << "keys jlA\n"; // turn off perspective and light, +anti-aliasing
      }
-      os << "keys cmA\n"; // show colorbar and mesh, +anti-aliasing
-      // update value-range; keep mesh-extents fixed
-      os << "autoscale value\n";
+      out << "keys cmA\n";        // show colorbar and mesh, +anti-aliasing
+      out << "autoscale value\n"; // update value-range; keep mesh-extents fixed
   }
-   os << flush;
+   out << flush;
 }

 void ReferenceConfiguration(const Vector &x, Vector &y)
@@ -34,43 +34,6 @@
 using namespace std;
 using namespace mfem;

-MatrixConstantCoefficient AnisotropicCoefficient(int dim, double anisotropy)
-{
-   DenseMatrix coeff(dim, dim);
-   coeff = 0.0;
-   coeff(0,0) = anisotropy;
-   for (int d = 1; d < dim; ++d)
-   {
-      coeff(d, d) = 1.0;
-   }
-   return coeff;
-}
-
-class SymmetricILUSmoother : public Solver
-{
-   BlockILU ilu;
-   double alpha;
-
-public:
-   SymmetricILUSmoother(Operator &op, double alpha_)
-      : ilu(op), alpha(alpha_)
-   { }
-
-   void Mult(const Vector &b, Vector &x) const
-   {
-      ilu.Mult(b, x);
-      x *= alpha;
-   }
-
-   void MultTranspose(const Vector &b, Vector &x) const
-   {
-      ilu.Mult(b, x);
-      x *= alpha;
-   }
-
-   void SetOperator(const Operator &op) { }
-};
-
 // Class for constructing a multigrid preconditioner for the diffusion operator.
 // This example multigrid preconditioner class demonstrates the creation of the
 // diffusion bilinear forms and operators using partial assembly for all spaces
@@ -80,18 +43,13 @@ public:
 class DiffusionMultigrid : public GeometricMultigrid
 {
 private:
-   MatrixConstantCoefficient coeff;
-   bool use_ilu;
+   ConstantCoefficient one;

 public:
   // Constructs a diffusion multigrid for the given FiniteElementSpaceHierarchy
   // and the array of essential boundaries
-   DiffusionMultigrid(FiniteElementSpaceHierarchy& fespaces, Array<int>& ess_bdr,
-                      double anisotropy, bool use_ilu_)
-      : GeometricMultigrid(fespaces),
-        coeff(AnisotropicCoefficient(fespaces.GetFinestFESpace().GetMesh()->Dimension(),
-                                     anisotropy)),
-        use_ilu(use_ilu_)
+   DiffusionMultigrid(FiniteElementSpaceHierarchy& fespaces, Array<int>& ess_bdr)
+      : GeometricMultigrid(fespaces), one(1.0)
   {
      ConstructCoarseOperatorAndSolver(fespaces.GetFESpaceAtLevel(0), ess_bdr);

@@ -105,7 +63,8 @@ private:
   void ConstructBilinearForm(FiniteElementSpace& fespace, Array<int>& ess_bdr)
   {
      BilinearForm* form = new BilinearForm(&fespace);
-      form->AddDomainIntegrator(new DiffusionIntegrator(coeff));
+      form->SetAssemblyLevel(AssemblyLevel::PARTIAL);
+      form->AddDomainIntegrator(new DiffusionIntegrator(one));
      form->Assemble();
      bfs.Append(form);

@@ -119,13 +78,18 @@ private:
      ConstructBilinearForm(coarse_fespace, ess_bdr);

      OperatorPtr opr;
-      opr.SetType(Operator::MFEM_SPARSEMAT);
+      opr.SetType(Operator::ANY_TYPE);
      bfs.Last()->FormSystemMatrix(*essentialTrueDofs.Last(), opr);
      opr.SetOperatorOwner(false);

-      UMFPackSolver *coarse_solver = new UMFPackSolver(*opr.As<SparseMatrix>());
+      CGSolver* pcg = new CGSolver();
+      pcg->SetPrintLevel(-1);
+      pcg->SetMaxIter(200);
+      pcg->SetRelTol(sqrt(1e-4));
+      pcg->SetAbsTol(0.0);
+      pcg->SetOperator(*opr.Ptr());

-      AddLevel(opr.Ptr(), coarse_solver, false, true);
+      AddLevel(opr.Ptr(), pcg, true, true);
   }

   void ConstructOperatorAndSmoother(FiniteElementSpace& fespace,
@@ -134,25 +98,16 @@ private:
      ConstructBilinearForm(fespace, ess_bdr);

      OperatorPtr opr;
-      opr.SetType(Operator::MFEM_SPARSEMAT);
+      opr.SetType(Operator::ANY_TYPE);
      bfs.Last()->FormSystemMatrix(*essentialTrueDofs.Last(), opr);
      opr.SetOperatorOwner(false);

-      Solver *smoother;
+      Vector diag(fespace.GetTrueVSize());
+      bfs.Last()->AssembleDiagonal(diag);

-      if (use_ilu)
-      {
-         smoother = new SymmetricILUSmoother(*opr, 0.5);
-      }
-      else
-      {
-         Vector diag(fespace.GetTrueVSize());
-         bfs.Last()->AssembleDiagonal(diag);
-         smoother = new OperatorChebyshevSmoother(
-            *opr, diag, *essentialTrueDofs.Last(), 2);
-      }
-
-      AddLevel(opr.Ptr(), smoother, false, true);
+      Solver* smoother = new OperatorChebyshevSmoother(*opr, diag,
+                                                       *essentialTrueDofs.Last(), 2);
+      AddLevel(opr.Ptr(), smoother, true, true);
   }
 };

@@ -165,8 +120,6 @@ int main(int argc, char *argv[])
   int order_refinements = 2;
   const char *device_config = "cpu";
   bool visualization = true;
-   double anisotropy = 1.0;
-   bool use_ilu = false;

   OptionsParser args(argc, argv);
   args.AddOption(&mesh_file, "-m", "--mesh",
@@ -175,11 +128,8 @@ int main(int argc, char *argv[])
                  "Number of geometric refinements done prior to order refinements.");
   args.AddOption(&order_refinements, "-or", "--order-refinements",
                  "Number of order refinements. Finest level in the hierarchy has order 2^{or}.");
-   args.AddOption(&anisotropy, "-a", "--anisotropy", "Anisotropy coefficient.");
   args.AddOption(&device_config, "-d", "--device",
                  "Device configuration string, see Device::Configure().");
-   args.AddOption(&use_ilu, "-i", "--use-ilu", "-no-i", "--no-ilu",
-                  "Use ILU smoothing?");
   args.AddOption(&visualization, "-vis", "--visualization", "-no-vis",
                  "--no-visualization",
                  "Enable or disable GLVis visualization.");
@@ -260,7 +210,7 @@ int main(int argc, char *argv[])
   Array<int> ess_bdr(mesh->bdr_attributes.Max());
   ess_bdr = 1;

-   DiffusionMultigrid M(fespaces, ess_bdr, anisotropy, use_ilu);
+   DiffusionMultigrid M(fespaces, ess_bdr);
   M.SetCycleType(Multigrid::CycleType::VCYCLE, 1, 1);

   OperatorPtr A;
@@ -75,7 +75,7 @@ Mesh * GenerateSerialMesh(int ref);
 // alpha*n.Grad(sol) + beta*sol - gamma over the same boundary.
 double IntegrateBC(const GridFunction &sol, const Array<int> &bdr_marker,
                   double alpha, double beta, double gamma,
-                   double &error);
+                   double &err);

 int main(int argc, char *argv[])
 {
@@ -295,33 +295,43 @@ int main(int argc, char *argv[])
   //     element solution.
   a.RecoverFEMSolution(X, b, u);

-   // 13. Compute the various boundary integrals.
+   // 13. Build a mass matrix to help solve for n.Grad(u) where 'n' is a surface
+   //     normal.
+   BilinearForm m(&fespace);
+   m.AddDomainIntegrator(new MassIntegrator);
+   m.Assemble();
+
+   ess_tdof_list.SetSize(0);
+   OperatorPtr M;
+   m.FormSystemMatrix(ess_tdof_list, M);
+
+   // 14. Compute the various boundary integrals.
   mfem::out << endl
             << "Verifying boundary conditions" << endl
             << "=============================" << endl;
   {
      // Integrate the solution on the Dirichlet boundary and compare to the
      // expected value.
-      double error, avg = IntegrateBC(u, dbc_bdr, 0.0, 1.0, dbc_val, error);
+      double err, avg = IntegrateBC(u, dbc_bdr, 0.0, 1.0, dbc_val, err);

      bool hom_dbc = (dbc_val == 0.0);
-      error /=  hom_dbc ? 1.0 : fabs(dbc_val);
+      err /=  hom_dbc ? 1.0 : fabs(dbc_val);
      mfem::out << "Average of solution on Gamma_dbc:\t"
                << avg << ", \t"
                << (hom_dbc ? "absolute" : "relative")
-                << " error " << error << endl;
+                << " error " << err << endl;
   }
   {
      // Integrate n.Grad(u) on the inhomogeneous Neumann boundary and compare
      // to the expected value.
-      double error, avg = IntegrateBC(u, nbc_bdr, 1.0, 0.0, nbc_val, error);
+      double err, avg = IntegrateBC(u, nbc_bdr, 1.0, 0.0, nbc_val, err);

      bool hom_nbc = (nbc_val == 0.0);
-      error /=  hom_nbc ? 1.0 : fabs(nbc_val);
+      err /=  hom_nbc ? 1.0 : fabs(nbc_val);
      mfem::out << "Average of n.Grad(u) on Gamma_nbc:\t"
                << avg << ", \t"
                << (hom_nbc ? "absolute" : "relative")
-                << " error " << error << endl;
+                << " error " << err << endl;
   }
   {
      // Integrate n.Grad(u) on the homogeneous Neumann boundary and compare to
@@ -330,29 +340,28 @@ int main(int argc, char *argv[])
      nbc0_bdr = 0;
      nbc0_bdr[3] = 1;

-      double error, avg = IntegrateBC(u, nbc0_bdr, 1.0, 0.0, 0.0, error);
+      double err, avg = IntegrateBC(u, nbc0_bdr, 1.0, 0.0, 0.0, err);

      bool hom_nbc = true;
      mfem::out << "Average of n.Grad(u) on Gamma_nbc0:\t"
                << avg << ", \t"
                << (hom_nbc ? "absolute" : "relative")
-                << " error " << error << endl;
+                << " error " << err << endl;
   }
   {
      // Integrate n.Grad(u) + a * u on the Robin boundary and compare to the
      // expected value.
-      double error;
-      double avg = IntegrateBC(u, rbc_bdr, 1.0, rbc_a_val, rbc_b_val, error);
+      double err, avg = IntegrateBC(u, rbc_bdr, 1.0, rbc_a_val, rbc_b_val, err);

      bool hom_rbc = (rbc_b_val == 0.0);
-      error /=  hom_rbc ? 1.0 : fabs(rbc_b_val);
+      err /=  hom_rbc ? 1.0 : fabs(rbc_b_val);
      mfem::out << "Average of n.Grad(u)+a*u on Gamma_rbc:\t"
                << avg << ", \t"
                << (hom_rbc ? "absolute" : "relative")
-                << " error " << error << endl;
+                << " error " << err << endl;
   }

-   // 14. Save the refined mesh and the solution. This output can be viewed
+   // 15. Save the refined mesh and the solution. This output can be viewed
   //     later using GLVis: "glvis -m refined.mesh -g sol.gf".
   {
      ofstream mesh_ofs("refined.mesh");
@@ -363,7 +372,7 @@ int main(int argc, char *argv[])
      u.Save(sol_ofs);
   }

-   // 15. Send the solution by socket to a GLVis server.
+   // 16. Send the solution by socket to a GLVis server.
   if (visualization)
   {
      string title_str = h1 ? "H1" : "DG";
@@ -376,7 +385,7 @@ int main(int argc, char *argv[])
               << " keys 'mmc'" << flush;
   }

-   // 16. Free the used memory.
+   // 17. Free the used memory.
   delete fec;
   delete mesh;

@@ -638,11 +647,11 @@ Mesh * GenerateSerialMesh(int ref)

 double IntegrateBC(const GridFunction &x, const Array<int> &bdr,
                   double alpha, double beta, double gamma,
-                   double &error)
+                   double &err)
 {
   double nrm = 0.0;
   double avg = 0.0;
-   error = 0.0;
+   err = 0.0;

   const bool a_is_zero = alpha == 0.0;
   const bool b_is_zero = beta == 0.0;
@@ -706,20 +715,20 @@ double IntegrateBC(const GridFunction &x, const Array<int> &bdr,

         // Integrate |alpha * n.Grad(x) + beta * x - gamma|^2
         val -= gamma;
-         error += (val*val) * ip.weight * face_weight;
+         err += (val*val) * ip.weight * face_weight;
      }
   }

   // Normalize by the length of the boundary
   if (std::abs(nrm) > 0.0)
   {
-      error /= nrm;
+      err /= nrm;
      avg /= nrm;
   }

   // Compute l2 norm of the error in the boundary condition (negative
-   // quadrature weights may produce negative 'error')
-   error = (error >= 0.0) ? sqrt(error) : -sqrt(-error);
+   // quadrature weights may produce negative 'err')
+   err = (err >= 0.0) ? sqrt(err) : -sqrt(-err);

   // Return the average value of alpha * n.Grad(x) + beta * x
   return avg;
@@ -75,7 +75,7 @@ Mesh * GenerateSerialMesh(int ref);
 // alpha*n.Grad(sol) + beta*sol - gamma over the same boundary.
 double IntegrateBC(const ParGridFunction &sol, const Array<int> &bdr_marker,
                   double alpha, double beta, double gamma,
-                   double &error);
+                   double &err);

 int main(int argc, char *argv[])
 {
@@ -314,33 +314,43 @@ int main(int argc, char *argv[])
   //     local finite element solution on each processor.
   a.RecoverFEMSolution(X, b, u);

-   // 14. Compute the various boundary integrals.
+   // 14. Build a mass matrix to help solve for n.Grad(u) where 'n' is a surface
+   //     normal.
+   ParBilinearForm m(&fespace);
+   m.AddDomainIntegrator(new MassIntegrator);
+   m.Assemble();
+
+   ess_tdof_list.SetSize(0);
+   OperatorPtr M;
+   m.FormSystemMatrix(ess_tdof_list, M);
+
+   // 15. Compute the various boundary integrals.
   mfem::out << endl
             << "Verifying boundary conditions" << endl
             << "=============================" << endl;
   {
      // Integrate the solution on the Dirichlet boundary and compare to the
      // expected value.
-      double error, avg = IntegrateBC(u, dbc_bdr, 0.0, 1.0, dbc_val, error);
+      double err, avg = IntegrateBC(u, dbc_bdr, 0.0, 1.0, dbc_val, err);

      bool hom_dbc = (dbc_val == 0.0);
-      error /=  hom_dbc ? 1.0 : fabs(dbc_val);
+      err /=  hom_dbc ? 1.0 : fabs(dbc_val);
      mfem::out << "Average of solution on Gamma_dbc:\t"
                << avg << ", \t"
                << (hom_dbc ? "absolute" : "relative")
-                << " error " << error << endl;
+                << " error " << err << endl;
   }
   {
      // Integrate n.Grad(u) on the inhomogeneous Neumann boundary and compare
      // to the expected value.
-      double error, avg = IntegrateBC(u, nbc_bdr, 1.0, 0.0, nbc_val, error);
+      double err, avg = IntegrateBC(u, nbc_bdr, 1.0, 0.0, nbc_val, err);

      bool hom_nbc = (nbc_val == 0.0);
-      error /=  hom_nbc ? 1.0 : fabs(nbc_val);
+      err /=  hom_nbc ? 1.0 : fabs(nbc_val);
      mfem::out << "Average of n.Grad(u) on Gamma_nbc:\t"
                << avg << ", \t"
                << (hom_nbc ? "absolute" : "relative")
-                << " error " << error << endl;
+                << " error " << err << endl;
   }
   {
      // Integrate n.Grad(u) on the homogeneous Neumann boundary and compare to
@@ -349,29 +359,28 @@ int main(int argc, char *argv[])
      nbc0_bdr = 0;
      nbc0_bdr[3] = 1;

-      double error, avg = IntegrateBC(u, nbc0_bdr, 1.0, 0.0, 0.0, error);
+      double err, avg = IntegrateBC(u, nbc0_bdr, 1.0, 0.0, 0.0, err);

      bool hom_nbc = true;
      mfem::out << "Average of n.Grad(u) on Gamma_nbc0:\t"
                << avg << ", \t"
                << (hom_nbc ? "absolute" : "relative")
-                << " error " << error << endl;
+                << " error " << err << endl;
   }
   {
      // Integrate n.Grad(u) + a * u on the Robin boundary and compare to the
      // expected value.
-      double error, avg = IntegrateBC(u, rbc_bdr, 1.0, rbc_a_val, rbc_b_val,
-                                      error);
+      double err, avg = IntegrateBC(u, rbc_bdr, 1.0, rbc_a_val, rbc_b_val, err);

      bool hom_rbc = (rbc_b_val == 0.0);
-      error /=  hom_rbc ? 1.0 : fabs(rbc_b_val);
+      err /=  hom_rbc ? 1.0 : fabs(rbc_b_val);
      mfem::out << "Average of n.Grad(u)+a*u on Gamma_rbc:\t"
                << avg << ", \t"
                << (hom_rbc ? "absolute" : "relative")
-                << " error " << error << endl;
+                << " error " << err << endl;
   }

-   // 15. Save the refined mesh and the solution in parallel. This output can be
+   // 16. Save the refined mesh and the solution in parallel. This output can be
   //     viewed later using GLVis: "glvis -np <np> -m mesh -g sol".
   {
      ostringstream mesh_name, sol_name;
@@ -387,7 +396,7 @@ int main(int argc, char *argv[])
      u.Save(sol_ofs);
   }

-   // 16. Send the solution by socket to a GLVis server.
+   // 17. Send the solution by socket to a GLVis server.
   if (visualization)
   {
      string title_str = h1 ? "H1" : "DG";
@@ -402,7 +411,7 @@ int main(int argc, char *argv[])
               << " keys 'mmc'" << flush;
   }

-   // 17. Free the used memory.
+   // 18. Free the used memory.
   delete fec;

   return 0;
@@ -668,11 +677,11 @@ double IntegrateBC(const ParGridFunction &x, const Array<int> &bdr,
   double loc_vals[3];
   double &nrm = loc_vals[0];
   double &avg = loc_vals[1];
-   double &error = loc_vals[2];
+   double &err = loc_vals[2];

   nrm = 0.0;
   avg = 0.0;
-   error = 0.0;
+   err = 0.0;

   const bool a_is_zero = alpha == 0.0;
   const bool b_is_zero = beta == 0.0;
@@ -736,7 +745,7 @@ double IntegrateBC(const ParGridFunction &x, const Array<int> &bdr,

         // Integrate |alpha * n.Grad(x) + beta * x - gamma|^2
         val -= gamma;
-         error += (val*val) * ip.weight * face_weight;
+         err += (val*val) * ip.weight * face_weight;
      }
   }

@@ -755,7 +764,7 @@ double IntegrateBC(const ParGridFunction &x, const Array<int> &bdr,
   }

   // Compute l2 norm of the error in the boundary condition (negative
-   // quadrature weights may produce negative 'error')
+   // quadrature weights may produce negative 'err')
   glb_err = (glb_err >= 0.0) ? sqrt(glb_err) : -sqrt(-glb_err);

   // Return the average value of alpha * n.Grad(x) + beta * x
@@ -81,10 +81,10 @@ Mesh * build_trapezoid_mesh(double offset)

 int main(int argc, char *argv[])
 {
-#ifdef HYPRE_USING_GPU
+#ifdef HYPRE_USING_CUDA
   cout << "\nAs of mfem-4.3 and hypre-2.22.0 (July 2021) this example\n"
-        << "is NOT supported with the GPU version of hypre.\n\n";
-   return 242;
+        << "is NOT supported with the CUDA version of hypre.\n\n";
+   return 255;
 #endif

   // 1. Initialize MPI.
@@ -167,9 +167,9 @@ int main(int argc, char *argv[])

   // 13. Compute error in the solution and its flux
   FunctionCoefficient uCoef(uExact);
-   double error = x.ComputeL2Error(uCoef);
+   double err = x.ComputeL2Error(uCoef);

-   cout << "|u - u_h|_2 = " << error << endl;
+   cout << "|u - u_h|_2 = " << err << endl;

   FiniteElementSpace flux_fespace(mesh, &fec, 3);
   GridFunction flux(&flux_fespace);
@@ -197,9 +197,9 @@ int main(int argc, char *argv[])

   // 15. Compute error in the solution and its flux
   FunctionCoefficient uCoef(uExact);
-   double error = x.ComputeL2Error(uCoef);
+   double err = x.ComputeL2Error(uCoef);

-   if (myid == 0) { cout << "|u - u_h|_2 = " << error << endl; }
+   if (myid == 0) { cout << "|u - u_h|_2 = " << err << endl; }

   ParFiniteElementSpace flux_fespace(&pmesh, &fec, 3);
   ParGridFunction flux(&flux_fespace);
@@ -1,195 +0,0 @@
-//                            MFEM Example 30
-//
-// Compile with: make ex30
-//
-// Sample runs:  ex30 -m ../data/square-disc.mesh -o 1
-//               ex30 -m ../data/square-disc.mesh -o 2
-//               ex30 -m ../data/square-disc.mesh -o 2 -me 1e3
-//               ex30 -m ../data/square-disc-nurbs.mesh -o 2
-//               ex30 -m ../data/star.mesh -o 2 -eo 4
-//               ex30 -m ../data/fichera.mesh -o 2 -me 1e4
-//               ex30 -m ../data/disc-nurbs.mesh -o 2
-//               ex30 -m ../data/ball-nurbs.mesh -o 2 -eo 3 -e 1e-2 -me 1e4
-//               ex30 -m ../data/star-surf.mesh -o 2
-//               ex30 -m ../data/square-disc-surf.mesh -o 2
-//               ex30 -m ../data/amr-quad.mesh -l 2
-//
-// Description:  This is an example of adaptive mesh refinement preprocessing
-//               which lowers the data oscillation [1] to a user-defined
-//               relative threshold. There is no PDE being solved.
-//
-//               MFEM's capability to work with both conforming and
-//               nonconforming meshes is demonstrated in example 6. In some
-//               problems, the material data or loading data is not sufficiently
-//               resolved on the initial mesh. This missing fine scale data
-//               reduces the accuracy of the solution as well as the accuracy
-//               of some local error estimators. By preprocessing the mesh
-//               before solving the PDE, many issues can be avoided.
-//
-//               [1] Morin, P., Nochetto, R. H., & Siebert, K. G. (2000).
-//                   Data oscillation and convergence of adaptive FEM. SIAM
-//                   Journal on Numerical Analysis, 38(2), 466-488.
-//
-//               [2] Mitchell, W. F. (2013). A collection of 2D elliptic
-//                   problems for testing adaptive grid refinement algorithms.
-//                   Applied mathematics and computation, 220, 350-364.
-
-#include "mfem.hpp"
-#include <fstream>
-#include <iostream>
-
-using namespace std;
-using namespace mfem;
-
-// Piecewise-affine function which is sometimes mesh-conforming
-double affine_function(const Vector &p)
-{
-   double x = p(0), y = p(1);
-   if (x < 0.0)
-   {
-      return 1.0 + x + y;
-   }
-   else
-   {
-      return 1.0;
-   }
-}
-
-// Piecewise-constant function which is never mesh-conforming
-double jump_function(const Vector &p)
-{
-   if (p.Normlp(2.0) > 0.4 && p.Normlp(2.0) < 0.6) { return 1.0; }
-   return 5.0;
-}
-
-// Singular function derived from the Laplacian of the "steep wavefront"
-// problem in [2].
-double singular_function(const Vector &p)
-{
-   double x = p(0), y = p(1);
-   double alpha = 1000.0;
-   double xc = 0.75, yc = 0.5;
-   double r0 = 0.7;
-   double r = sqrt(pow(x - xc,2.0) + pow(y - yc,2.0));
-   double num = - ( alpha - pow(alpha,3) * (pow(r,2) - pow(r0,2)) );
-   double denom = pow(r * ( pow(alpha,2) * pow(r0,2) + pow(alpha,2) * pow(r,2) \
-                            - 2 * pow(alpha,2) * r0 * r + 1.0 ),2);
-   denom = max(denom,1e-8);
-   return num / denom;
-}
-
-int main(int argc, char *argv[])
-{
-   // 1. Parse command-line options.
-   const char *mesh_file = "../data/star.mesh";
-   int order = 1;
-   int nc_limit = 1;
-   int max_elems = 1e5;
-   double double_max_elems = double(max_elems);
-   bool visualization = true;
-   double osc_threshold = 1e-3;
-   int enriched_order = 5;
-
-   OptionsParser args(argc, argv);
-   args.AddOption(&mesh_file, "-m", "--mesh",
-                  "Mesh file to use.");
-   args.AddOption(&order, "-o", "--order",
-                  "Finite element order (polynomial degree).");
-   args.AddOption(&nc_limit, "-l", "--nc-limit",
-                  "Maximum level of hanging nodes.");
-   args.AddOption(&double_max_elems, "-me", "--max-elems",
-                  "Stop after reaching this many elements.");
-   args.AddOption(&visualization, "-vis", "--visualization", "-no-vis",
-                  "--no-visualization",
-                  "Enable or disable GLVis visualization.");
-   args.AddOption(&osc_threshold, "-e", "--error",
-                  "relative data oscillation threshold.");
-   args.AddOption(&enriched_order, "-eo", "--enriched_order",
-                  "Enriched quadrature order.");
-
-   args.Parse();
-   if (!args.Good())
-   {
-      args.PrintUsage(cout);
-      return 1;
-   }
-   args.PrintOptions(cout);
-
-   max_elems = int(double_max_elems);
-   Mesh mesh(mesh_file, 1, 1);
-
-   // 2. Since a NURBS mesh can currently only be refined uniformly, we need to
-   //    convert it to a piecewise-polynomial curved mesh. First we refine the
-   //    NURBS mesh a bit more and then project the curvature to quadratic Nodes.
-   if (mesh.NURBSext)
-   {
-      for (int i = 0; i < 2; i++)
-      {
-         mesh.UniformRefinement();
-      }
-      mesh.SetCurvature(2);
-   }
-
-   // 3. Define functions and refiner.
-   FunctionCoefficient affine_coeff(affine_function);
-   FunctionCoefficient jump_coeff(jump_function);
-   FunctionCoefficient singular_coeff(singular_function);
-   CoefficientRefiner coeffrefiner(affine_coeff, order);
-
-   // 4. Connect to GLVis.
-   char vishost[] = "localhost";
-   int  visport   = 19916;
-   socketstream sol_sock;
-   if (visualization)
-   {
-      sol_sock.open(vishost, visport);
-   }
-
-   // 5. Define custom integration rule (optional).
-   const IntegrationRule *irs[Geometry::NumGeom];
-   int order_quad = 2*order + enriched_order;
-   for (int i = 0; i < Geometry::NumGeom; ++i)
-   {
-      irs[i] = &(IntRules.Get(i, order_quad));
-   }
-
-   // 6. Apply custom refiner settings.
-   coeffrefiner.SetIntRule(irs);
-   coeffrefiner.SetMaxElements(max_elems);
-   coeffrefiner.SetThreshold(osc_threshold);
-   coeffrefiner.SetNCLimit(nc_limit);
-   coeffrefiner.PrintWarnings();
-
-   // 7. Preprocess mesh to control osc (piecewise-affine function).
-   //    This is mostly just a verification check. The oscillation should
-   //    be zero if the function is mesh-conforming and order > 0.
-   coeffrefiner.PreprocessMesh(mesh);
-
-   mfem::out  << "\n";
-   mfem::out << "Function 0 (affine) \n";
-   mfem::out << "Number of Elements " << mesh.GetNE() << "\n";
-   mfem::out << "Osc error " << coeffrefiner.GetOsc() << "\n";
-
-   // 8. Preprocess mesh to control osc (jump function).
-   coeffrefiner.ResetCoefficient(jump_coeff);
-   coeffrefiner.PreprocessMesh(mesh);
-
-   mfem::out  << "\n";
-   mfem::out << "Function 1 (discontinuous) \n";
-   mfem::out << "Number of Elements " << mesh.GetNE() << "\n";
-   mfem::out << "Osc error " << coeffrefiner.GetOsc() << "\n";
-
-   // 9. Preprocess mesh to control osc (singular function).
-   coeffrefiner.ResetCoefficient(singular_coeff);
-   coeffrefiner.PreprocessMesh(mesh);
-
-   mfem::out  << "\n";
-   mfem::out << "Function 2 (singular) \n";
-   mfem::out << "Number of Elements " << mesh.GetNE() << "\n";
-   mfem::out << "Osc error " << coeffrefiner.GetOsc() << "\n";
-
-   sol_sock.precision(8);
-   sol_sock << "mesh\n" << mesh << flush;
-
-   return 0;
-}
@@ -1,241 +0,0 @@
-//                        MFEM Example 30 - Parallel Version
-//
-// Compile with: make ex30p
-//
-// Sample runs:  mpirun -np 4 ex30p -m ../data/square-disc.mesh -o 1
-//               mpirun -np 4 ex30p -m ../data/square-disc.mesh -o 2
-//               mpirun -np 4 ex30p -m ../data/square-disc.mesh -o 2 -me 1e3
-//               mpirun -np 4 ex30p -m ../data/square-disc-nurbs.mesh -o 2
-//               mpirun -np 4 ex30p -m ../data/star.mesh -o 2 -eo 4
-//               mpirun -np 4 oscp -m ../data/fichera.mesh -o 2 -me 1e4
-//               mpirun -np 4 ex30p -m ../data/disc-nurbs.mesh -o 2
-//               mpirun -np 4 ex30p -m ../data/ball-nurbs.mesh -o 2 -eo 3 -e 1e-2
-//               mpirun -np 4 ex30p -m ../data/star-surf.mesh -o 2
-//               mpirun -np 4 ex30p -m ../data/square-disc-surf.mesh -o 2
-//               mpirun -np 4 ex30p -m ../data/amr-quad.mesh -l 2
-//
-// Description:  This is an example of adaptive mesh refinement preprocessing
-//               which lowers the data oscillation [1] to a user-defined
-//               relative threshold. There is no PDE being solved.
-//
-//               MFEM's capability to work with both conforming and
-//               nonconforming meshes is demonstrated in example 6. In some
-//               problems, the material data or loading data is not sufficiently
-//               resolved on the initial mesh. This missing fine scale data
-//               reduces the accuracy of the solution as well as the accuracy
-//               of some local error estimators. By preprocessing the mesh
-//               before solving the PDE, many issues can be avoided.
-//
-//               [1] Morin, P., Nochetto, R. H., & Siebert, K. G. (2000).
-//                   Data oscillation and convergence of adaptive FEM. SIAM
-//                   Journal on Numerical Analysis, 38(2), 466-488.
-//
-//               [2] Mitchell, W. F. (2013). A collection of 2D elliptic
-//                   problems for testing adaptive grid refinement algorithms.
-//                   Applied mathematics and computation, 220, 350-364.
-
-#include "mfem.hpp"
-#include <fstream>
-#include <iostream>
-
-using namespace std;
-using namespace mfem;
-
-// Piecewise-affine function which is sometimes mesh-conforming
-double affine_function(const Vector &p)
-{
-   double x = p(0), y = p(1);
-   if (x < 0.0)
-   {
-      return 1.0 + x + y;
-   }
-   else
-   {
-      return 1.0;
-   }
-}
-
-// Piecewise-constant function which is never mesh-conforming
-double jump_function(const Vector &p)
-{
-   if (p.Normlp(2.0) > 0.4 && p.Normlp(2.0) < 0.6) { return 1.0; }
-   return 5.0;
-}
-
-// Singular function derived from the Laplacian of the "steep wavefront"
-// problem in [2].
-double singular_function(const Vector &p)
-{
-   double x = p(0), y = p(1);
-   double alpha = 1000.0;
-   double xc = 0.75, yc = 0.5;
-   double r0 = 0.7;
-   double r = sqrt(pow(x - xc,2.0) + pow(y - yc,2.0));
-   double num = - ( alpha - pow(alpha,3) * (pow(r,2) - pow(r0,2)) );
-   double denom = pow(r * ( pow(alpha,2) * pow(r0,2) + pow(alpha,2) * pow(r,2) \
-                            - 2 * pow(alpha,2) * r0 * r + 1.0 ),2);
-   denom = max(denom,1e-8);
-   return num / denom;
-}
-
-int main(int argc, char *argv[])
-{
-   // 0. Initialize MPI.
-   int num_procs, myid;
-   MPI_Init(&argc, &argv);
-   MPI_Comm_size(MPI_COMM_WORLD, &num_procs);
-   MPI_Comm_rank(MPI_COMM_WORLD, &myid);
-
-   // 1. Parse command-line options.
-   const char *mesh_file = "../data/star.mesh";
-   int order = 1;
-   int nc_limit = 1;
-   int max_elems = 1e5;
-   double double_max_elems = double(max_elems);
-   bool visualization = true;
-   bool nc_simplices = true;
-   double osc_threshold = 1e-3;
-   int enriched_order = 5;
-
-   OptionsParser args(argc, argv);
-   args.AddOption(&mesh_file, "-m", "--mesh",
-                  "Mesh file to use.");
-   args.AddOption(&order, "-o", "--order",
-                  "Finite element order (polynomial degree).");
-   args.AddOption(&nc_limit, "-l", "--nc-limit",
-                  "Maximum level of hanging nodes.");
-   args.AddOption(&double_max_elems, "-me", "--max-elems",
-                  "Stop after reaching this many elements.");
-   args.AddOption(&visualization, "-vis", "--visualization", "-no-vis",
-                  "--no-visualization",
-                  "Enable or disable GLVis visualization.");
-   args.AddOption(&osc_threshold, "-e", "--error",
-                  "relative data oscillation threshold.");
-   args.AddOption(&enriched_order, "-eo", "--enriched_order",
-                  "Enriched quadrature order.");
-   args.AddOption(&nc_simplices, "-ns", "--nonconforming-simplices",
-                  "-cs", "--conforming-simplices",
-                  "For simplicial meshes, enable/disable nonconforming"
-                  " refinement");
-
-   args.Parse();
-   if (!args.Good())
-   {
-      if (myid == 0)
-      {
-         args.PrintUsage(cout);
-      }
-      MPI_Finalize();
-      return 1;
-   }
-   if (myid == 0)
-   {
-      args.PrintOptions(cout);
-   }
-
-   max_elems = int(double_max_elems);
-   Mesh mesh(mesh_file, 1, 1);
-
-   // 2. Since a NURBS mesh can currently only be refined uniformly, we need to
-   //    convert it to a piecewise-polynomial curved mesh. First we refine the
-   //    NURBS mesh a bit more and then project the curvature to quadratic Nodes.
-   if (mesh.NURBSext)
-   {
-      for (int i = 0; i < 2; i++)
-      {
-         mesh.UniformRefinement();
-      }
-      mesh.SetCurvature(2);
-   }
-
-   // 3. Make sure the mesh is in the non-conforming mode to enable local
-   //    refinement of quadrilaterals/hexahedra. Simplices can be refined
-   //    either in conforming or in non-conforming mode. The conforming
-   //    mode however does not support dynamic partitioning.
-   mesh.EnsureNCMesh(nc_simplices);
-
-   // 4. Define a parallel mesh by partitioning the serial mesh.
-   //    Once the parallel mesh is defined, the serial mesh can be deleted.
-   ParMesh pmesh(MPI_COMM_WORLD, mesh);
-   mesh.Clear();
-
-   // 5. Define functions and refiner.
-   FunctionCoefficient affine_coeff(affine_function);
-   FunctionCoefficient jump_coeff(jump_function);
-   FunctionCoefficient singular_coeff(singular_function);
-   CoefficientRefiner coeffrefiner(affine_coeff,order);
-
-   // 6. Connect to GLVis.
-   char vishost[] = "localhost";
-   int  visport   = 19916;
-   socketstream sol_sock;
-   if (visualization)
-   {
-      sol_sock.open(vishost, visport);
-   }
-
-   // 7. Define custom integration rule (optional).
-   const IntegrationRule *irs[Geometry::NumGeom];
-   int order_quad = 2*order + enriched_order;
-   for (int i=0; i < Geometry::NumGeom; ++i)
-   {
-      irs[i] = &(IntRules.Get(i, order_quad));
-   }
-
-   // 8. Apply custom refiner settings.
-   coeffrefiner.SetIntRule(irs);
-   coeffrefiner.SetMaxElements(max_elems);
-   coeffrefiner.SetThreshold(osc_threshold);
-   coeffrefiner.SetNCLimit(nc_limit);
-   coeffrefiner.PrintWarnings();
-
-   // 9. Preprocess mesh to control osc (piecewise-affine function).
-   //    This is mostly just a verification check. The oscillation should
-   //    be zero if the function is mesh-conforming and order > 0.
-   coeffrefiner.PreprocessMesh(pmesh);
-
-   int globalNE = pmesh.GetGlobalNE();
-   double osc = coeffrefiner.GetOsc();
-   if (myid == 0)
-   {
-      mfem::out  << "\n";
-      mfem::out << "Function 0 (affine) \n";
-      mfem::out << "Number of Elements " << globalNE << "\n";
-      mfem::out << "Osc error " << osc << "\n";
-   }
-
-   // 10. Preprocess mesh to control osc (jump function).
-   coeffrefiner.ResetCoefficient(jump_coeff);
-   coeffrefiner.PreprocessMesh(pmesh);
-
-   globalNE = pmesh.GetGlobalNE();
-   osc = coeffrefiner.GetOsc();
-   if (myid == 0)
-   {
-      mfem::out  << "\n";
-      mfem::out << "Function 1 (discontinuous) \n";
-      mfem::out << "Number of Elements " << globalNE << "\n";
-      mfem::out << "Osc error " << osc << "\n";
-   }
-
-   // 11. Preprocess mesh to control osc (singular function).
-   coeffrefiner.ResetCoefficient(singular_coeff);
-   coeffrefiner.PreprocessMesh(pmesh);
-
-   globalNE = pmesh.GetGlobalNE();
-   osc = coeffrefiner.GetOsc();
-   if (myid == 0)
-   {
-      mfem::out  << "\n";
-      mfem::out << "Function 2 (singular) \n";
-      mfem::out << "Number of Elements " << globalNE << "\n";
-      mfem::out << "Osc error " << osc << "\n";
-   }
-
-   sol_sock.precision(8);
-   sol_sock << "parallel " << num_procs << " " << myid << "\n";
-   sol_sock << "mesh\n" << pmesh << flush;
-
-   MPI_Finalize();
-   return 0;
-}
@@ -255,10 +255,10 @@ int main(int argc, char *argv[])

   // 15. Compute and print the L^2 norm of the error.
   {
-      double error = x.ComputeL2Error(E);
+      double err = x.ComputeL2Error(E);
      if (myid == 0)
      {
-         cout << "\n|| E_h - E ||_{L^2} = " << error << '\n' << endl;
+         cout << "\n|| E_h - E ||_{L^2} = " << err << '\n' << endl;
      }
   }

@@ -256,10 +256,10 @@ int main(int argc, char *argv[])

   // 15. Compute and print the L^2 norm of the error.
   {
-      double error = x.ComputeL2Error(F);
+      double err = x.ComputeL2Error(F);
      if (myid == 0)
      {
-         cout << "\n|| F_h - F ||_{L^2} = " << error << '\n' << endl;
+         cout << "\n|| F_h - F ||_{L^2} = " << err << '\n' << endl;
      }
   }

@@ -282,10 +282,10 @@ int main(int argc, char *argv[])
   delete b;

   // 12. Compute and print the L^2 norm of the error.
-   double error = x.ComputeL2Error(sol_coef);
+   double err = x.ComputeL2Error(sol_coef);
   if (myid == 0)
   {
-      cout << "\nL2 norm of error: " << error << endl;
+      cout << "\nL2 norm of error: " << err << endl;
   }

   // 13. Save the refined mesh and the solution. This output can be viewed
@@ -31,31 +31,29 @@ add_mfem_examples(GINKGO_EXAMPLES_SRCS ${PFX} "" test_ginkgo)
 # which builds the examples and runs:
 #   ctest -R ginkgo

-if (MFEM_ENABLE_TESTING)
-  # Command line options for the tests.
-  set(EX1_COMMON_OPTS ex1 -m ../data/star.mesh --use_gko_solver)
-  set(EX1_TEST_OPTS ${EX9_COMMON_OPTS})
+# Command line options for the tests.
+set(EX1_COMMON_OPTS ex1 -m ../data/star.mesh --use_gko_solver)
+set(EX1_TEST_OPTS ${EX9_COMMON_OPTS})

-  # Add the tests: one test per source file.
-  foreach(SRC_FILE ${GINKGO_EXAMPLES_SRCS})
-    get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
-    string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
-    string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
-    set(TEST_NAME ${PFX}${TEST_NAME})
+# Add the tests: one test per source file.
+foreach(SRC_FILE ${GINKGO_EXAMPLES_SRCS})
+  get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
+  string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
+  string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
+  set(TEST_NAME ${PFX}${TEST_NAME})

-    set(THIS_TEST_OPTIONS "-no-vis")
-    list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
-    # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")
+  set(THIS_TEST_OPTIONS "-no-vis")
+  list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
+  # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")

-    if (NOT (${TEST_NAME} MATCHES ".*p$"))
-      add_test(NAME ${TEST_NAME}_ser
-        COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
-    else()
-      add_test(NAME ${TEST_NAME}_np=4
-        COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
-        ${MPIEXEC_PREFLAGS}
-        $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
-        ${MPIEXEC_POSTFLAGS})
-    endif()
-  endforeach()
-endif()
+  if (NOT (${TEST_NAME} MATCHES ".*p$"))
+    add_test(NAME ${TEST_NAME}_ser
+      COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
+  else()
+    add_test(NAME ${TEST_NAME}_np=4
+      COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
+      ${MPIEXEC_PREFLAGS}
+      $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
+      ${MPIEXEC_POSTFLAGS})
+  endif()
+endforeach()
@@ -33,33 +33,31 @@ add_mfem_examples(HIOP_EXAMPLES_SRCS ${PFX} "" test_hiop)
 # which builds the examples and runs:
 #   ctest -R hiop

-if (MFEM_ENABLE_TESTING)
-  # Command line options for the tests.
-  # Example 9:
-  set(EX9_COMMON_OPTS -m ../../data/periodic-segment.mesh -p 0 -dt 0.005)
-  set(EX9_TEST_OPTS ${EX9_COMMON_OPTS}  -r  2 )
-  set(EX9P_TEST_OPTS ${EX9_COMMON_OPTS})
+# Command line options for the tests.
+# Example 9:
+set(EX9_COMMON_OPTS -m ../../data/periodic-segment.mesh -p 0 -dt 0.005)
+set(EX9_TEST_OPTS ${EX9_COMMON_OPTS}  -r  2 )
+set(EX9P_TEST_OPTS ${EX9_COMMON_OPTS})

-  # Add the tests: one test per source file.
-  foreach(SRC_FILE ${HIOP_EXAMPLES_SRCS})
-    get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
-    string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
-    string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
-    set(TEST_NAME ${PFX}${TEST_NAME})
+# Add the tests: one test per source file.
+foreach(SRC_FILE ${HIOP_EXAMPLES_SRCS})
+  get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
+  string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
+  string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
+  set(TEST_NAME ${PFX}${TEST_NAME})

-    set(THIS_TEST_OPTIONS "-no-vis")
-    list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
-    # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")
+  set(THIS_TEST_OPTIONS "-no-vis")
+  list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
+  # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")

-    if (NOT (${TEST_NAME} MATCHES ".*p$"))
-      add_test(NAME ${TEST_NAME}_ser
-        COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
-    else()
-      add_test(NAME ${TEST_NAME}_np=4
-        COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} 4
-        ${MPIEXEC_PREFLAGS}
-        $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
-        ${MPIEXEC_POSTFLAGS})
-    endif()
-  endforeach()
-endif()
+  if (NOT (${TEST_NAME} MATCHES ".*p$"))
+    add_test(NAME ${TEST_NAME}_ser
+      COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
+  else()
+    add_test(NAME ${TEST_NAME}_np=4
+      COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} 4
+      ${MPIEXEC_PREFLAGS}
+      $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
+      ${MPIEXEC_POSTFLAGS})
+  endif()
+endforeach()
@@ -1,31 +0,0 @@
-# Jupyter Notebooks using xeus-cling
-
-[![Binder](https://mybinder.org/badge_logo.svg)](https://mybinder.org/v2/gh/mfem/mfem/master?filepath=examples%2Fjupyter%2Fex.ipynb)
-
-[xeus-cling](https://github.com/jupyter-xeus/xeus-cling) is a C++ Jupyter Kernel based on [cling](https://github.com/root-project/cling),
-which can be used to create interactive C++ MFEM and GLVis notebooks.
-
-Click on the `binder` button above for an interactive example.
-
-## Installing Locally
-
-In order to run notebooks locally you will need `xeus-cling` along with `mfem` and `xglvis`. We recommend you use
-[miniconda](https://docs.conda.io/en/latest/miniconda.html) or, if you already have it installed,
-[conda](https://docs.conda.io/projects/conda/en/latest/).
-
-1. Follow the install steps on https://github.com/jupyter-xeus/xeus-cling to install the C++ kernels
-2. Build and install a _shared_ version of mfem
-    * for example: `make serial SHARED=YES`
-3. Install [pyglvis](https://github.com/glvis/pyglvis)
-    * for the widget frontend
-4. Get [xeus-glvis](https://github.com/glvis/xeus-glvis) and `cp` the header to `{PREFIX}/glvis/xglvis.hpp`
-    * (this could be improved)
-
-## Running Locally
-
-Once you've installed Jupyter, the C++ Kernel, mfem, and glvis start the notebook server (`jupyter-notebook`)
-and open an existing example or a new `C++ 1x` kernel.
-
-You will _always_ need to `#pragma cling load("mfem")` and you may need to point the `cling` runtime at your
-mfem and/or glvis installs, do this with the
-`#pragma cling` [statements](https://xeus-cling.readthedocs.io/en/latest/build_options.html#using-third-party-libraries).
@@ -1,155 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "markdown",
-   "id": "owned-extraction",
-   "metadata": {},
-   "source": [
-    "## Load the MFEM library\n",
-    "\n",
-    "Any non-default libraries must be loaded before you can `#include` files that use them. For more info see the [xeus-cling help](https://xeus-cling.readthedocs.io/en/latest/build_options.html)."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "waiting-portrait",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "#pragma cling load(\"mfem\")"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "foreign-recycling",
-   "metadata": {},
-   "source": [
-    "## MFEM Example 1"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "public-white",
-   "metadata": {},
-   "source": [
-    "This is the simplest MFEM example and a good starting point for new users. The example demonstrates the use of MFEM to define and solve an $H^1$ finite element discretization of the Laplace problem\n",
-    "\n",
-    "$$\n",
-    "-\\Delta u = 1\n",
-    "$$\n",
-    "\n",
-    "with homogeneous Dirichlet boundary conditions $u=0$.\n",
-    "\n",
-    "The example illustrates the use of the basic MFEM classes for defining the mesh, finite element space, as well as linear and bilinear forms corresponding to the left-hand side and right-hand side of the discrete linear system.\n",
-    "\n",
-    "Compare with MFEM's [ex1.cpp](https://github.com/mfem/mfem/blob/master/examples/ex1.cpp) and PyMFEM's [ex1.py](https://github.com/mfem/PyMFEM/blob/master/examples/ex1.py)."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "protective-darkness",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "#include <fstream>\n",
-    "#include <iostream>\n",
-    "#include <sstream>\n",
-    "\n",
-    "#include <mfem.hpp>\n",
-    "#include <glvis/xglvis.hpp>"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "falling-monkey",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "using namespace std;\n",
-    "using namespace mfem;\n",
-    "\n",
-    "Mesh mesh = Mesh::MakeCartesian2D(5, 5, Element::TRIANGLE);\n",
-    "mesh.UniformRefinement();\n",
-    "\n",
-    "H1_FECollection fec(2, mesh.Dimension());\n",
-    "\n",
-    "FiniteElementSpace fespace(&mesh, &fec);\n",
-    "cout << \"Number of finite element unknowns: \" << fespace.GetTrueVSize() << endl;\n",
-    "\n",
-    "Array<int> ess_tdof_list;\n",
-    "if (mesh.bdr_attributes.Size())\n",
-    "{\n",
-    "  Array<int> ess_bdr(mesh.bdr_attributes.Max());\n",
-    "  ess_bdr = 1;\n",
-    "  fespace.GetEssentialTrueDofs(ess_bdr, ess_tdof_list);\n",
-    "}\n",
-    "\n",
-    "LinearForm b(&fespace);\n",
-    "ConstantCoefficient one(1.0);\n",
-    "b.AddDomainIntegrator(new DomainLFIntegrator(one));\n",
-    "b.Assemble();\n",
-    "\n",
-    "GridFunction x(&fespace);\n",
-    "x = 0.0;\n",
-    "\n",
-    "BilinearForm a(&fespace);\n",
-    "a.AddDomainIntegrator(new DiffusionIntegrator(one));\n",
-    "a.Assemble();\n",
-    "\n",
-    "OperatorPtr A;\n",
-    "Vector B, X;\n",
-    "a.FormLinearSystem(ess_tdof_list, x, b, A, X, B);\n",
-    "\n",
-    "cout << \"Size of linear system: \" << A->Height() << endl;\n",
-    "\n",
-    "GSSmoother M((SparseMatrix&)(*A));\n",
-    "PCG(*A, M, B, X, 1, 200, 1e-12, 0.0);\n",
-    "a.RecoverFEMSolution(X, b, x);"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "hawaiian-republican",
-   "metadata": {},
-   "source": [
-    "## GLVis Visualization\n",
-    "\n",
-    "For now we save the computational mesh and finite element solution in a string and pass that to the glvis widget, see https://github.com/glvis/xeus-glvis for the widget backend and https://github.com/GLVis/pyglvis/tree/master/js for the widget frontend."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "ordinary-equation",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "std::stringstream ss;\n",
-    "ss << \"solution\\n\" << mesh << x << flush;\n",
-    "\n",
-    "auto glv = glvis::glvis();\n",
-    "glv.plot(ss.str() + \"keys Rjml\"); // the `+ \"keys ....\"' is optional\n",
-    "glv"
-   ]
-  }
- ],
- "metadata": {
-  "kernelspec": {
-   "display_name": "C++14",
-   "language": "C++14",
-   "name": "xcpp14"
-  },
-  "language_info": {
-   "codemirror_mode": "text/x-c++src",
-   "file_extension": ".cpp",
-   "mimetype": "text/x-c++src",
-   "name": "c++",
-   "version": "14"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 5
-}
@@ -22,10 +22,10 @@ MFEM_LIB_FILE = mfem_is_not_built
 -include $(CONFIG_MK)

 SEQ_EXAMPLES = ex0 ex1 ex2 ex3 ex4 ex5 ex6 ex7 ex8 ex9 ex10 ex14 ex15 ex16 \
-  ex17 ex18 ex19 ex20 ex21 ex22 ex23 ex24 ex25 ex26 ex27 ex28 ex29 ex30
+  ex17 ex18 ex19 ex20 ex21 ex22 ex23 ex24 ex25 ex26 ex27 ex28 ex29
 PAR_EXAMPLES = ex0p ex1p ex2p ex3p ex4p ex5p ex6p ex7p ex8p ex9p ex10p ex11p \
  ex12p ex13p ex14p ex15p ex16p ex17p ex18p ex19p ex20p ex21p ex22p ex24p \
-  ex25p ex26p ex27p ex28p ex29p ex30p
+  ex25p ex26p ex27p ex28p ex29p
 SEQ_DEVICE_EXAMPLES = ex1 ex3 ex4 ex5 ex6 ex9 ex22 ex24 ex25 ex26
 PAR_DEVICE_EXAMPLES = ex1p ex2p ex3p ex4p ex5p ex6p ex7p ex9p ex13p ex22p \
  ex24p ex25p ex26p
@@ -94,32 +94,30 @@ if (MFEM_USE_SLEPC)
 endif()

 # Add the tests: one test per command-line-variable.
-if (MFEM_ENABLE_TESTING)
-  set(TEST_OPTIONS_VARS
-      EX1_ARGS_W EX1_ARGS_P EX2_ARGS EX3_ARGS EX4_ARGS EX4_HYB_ARGS
-      EX5_BDDC_LB_ARGS EX5_BDDC_GB_ARGS EX5_FSPL_ARGS EX6_ARGS EX6_NONOVL_ARGS
-      EX9_E_ARGS EX9_ES_ARGS EX9_IS_ARGS EX10_ARGS)
-  if (MFEM_USE_SLEPC)
-      list(APPEND TEST_OPTIONS_VARS EX11_ARGS_SINV EX11_ARGS_LOBPCG EX11_ARGS_GD)
-  endif()
-
-  foreach(TEST_OPTIONS_VAR ${TEST_OPTIONS_VARS})
-    string(REGEX REPLACE "^(.+)_ARGS" "\\1" TEST_NAME_UC ${TEST_OPTIONS_VAR})
-    string(REGEX REPLACE "^([^_]+)" "\\1P" TEST_NAME_UC ${TEST_NAME_UC})
-    string(TOLOWER ${TEST_NAME_UC} TEST_NAME_FULL)
-    string(REGEX REPLACE "^([^_]+).*" "\\1" TEST_NAME ${TEST_NAME_FULL})
-    set(TEST_NAME_FULL ${PFX}${TEST_NAME_FULL})
-    set(TEST_NAME ${PFX}${TEST_NAME})
-    set(TEST_OPTIONS "-no-vis" ${${TEST_OPTIONS_VAR}})
-    # message(STATUS "${TEST_NAME_FULL} --> ${TEST_NAME} ${TEST_OPTIONS}")
-
-    # All PETSC tests are parallel.
-    if (MFEM_USE_MPI)
-      add_test(NAME ${TEST_NAME_FULL}_np=4
-        COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
-        ${MPIEXEC_PREFLAGS}
-        $<TARGET_FILE:${TEST_NAME}> ${TEST_OPTIONS}
-        ${MPIEXEC_POSTFLAGS})
-    endif()
-  endforeach()
+set(TEST_OPTIONS_VARS
+    EX1_ARGS_W EX1_ARGS_P EX2_ARGS EX3_ARGS EX4_ARGS EX4_HYB_ARGS
+    EX5_BDDC_LB_ARGS EX5_BDDC_GB_ARGS EX5_FSPL_ARGS EX6_ARGS EX6_NONOVL_ARGS
+    EX9_E_ARGS EX9_ES_ARGS EX9_IS_ARGS EX10_ARGS)
+if (MFEM_USE_SLEPC)
+    list(APPEND TEST_OPTIONS_VARS EX11_ARGS_SINV EX11_ARGS_LOBPCG EX11_ARGS_GD)
 endif()
+
+foreach(TEST_OPTIONS_VAR ${TEST_OPTIONS_VARS})
+  string(REGEX REPLACE "^(.+)_ARGS" "\\1" TEST_NAME_UC ${TEST_OPTIONS_VAR})
+  string(REGEX REPLACE "^([^_]+)" "\\1P" TEST_NAME_UC ${TEST_NAME_UC})
+  string(TOLOWER ${TEST_NAME_UC} TEST_NAME_FULL)
+  string(REGEX REPLACE "^([^_]+).*" "\\1" TEST_NAME ${TEST_NAME_FULL})
+  set(TEST_NAME_FULL ${PFX}${TEST_NAME_FULL})
+  set(TEST_NAME ${PFX}${TEST_NAME})
+  set(TEST_OPTIONS "-no-vis" ${${TEST_OPTIONS_VAR}})
+  # message(STATUS "${TEST_NAME_FULL} --> ${TEST_NAME} ${TEST_OPTIONS}")
+
+  # All PETSC tests are parallel.
+  if (MFEM_USE_MPI)
+    add_test(NAME ${TEST_NAME_FULL}_np=4
+      COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
+      ${MPIEXEC_PREFLAGS}
+      $<TARGET_FILE:${TEST_NAME}> ${TEST_OPTIONS}
+      ${MPIEXEC_POSTFLAGS})
+  endif()
+endforeach()
@@ -37,39 +37,37 @@ add_mfem_examples(PUMI_EXAMPLES_SRCS ${PFX} "" test_pumi)
 # which builds the examples and runs:
 #   ctest -R pumi

-if (MFEM_ENABLE_TESTING)
-  # Command line options for the tests.
-  # TODO...
+# Command line options for the tests.
+# TODO...

-  # Set the number of processors for the parallel examples. The value of
-  # MFEM_MPI_NP is ignored.
-  set(EX1_TEST_NP 1)
-  set(EX1P_TEST_NP 8)
-  set(EX2_TEST_NP 1)
-  set(EX6P_TEST_NP 8)
+# Set the number of processors for the parallel examples. The value of
+# MFEM_MPI_NP is ignored.
+set(EX1_TEST_NP 1)
+set(EX1P_TEST_NP 8)
+set(EX2_TEST_NP 1)
+set(EX6P_TEST_NP 8)

-  # Add the tests: one test per source file.
-  foreach(SRC_FILE ${PUMI_EXAMPLES_SRCS})
-    get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
-    string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
-    string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
-    set(TEST_NAME ${PFX}${TEST_NAME})
+# Add the tests: one test per source file.
+foreach(SRC_FILE ${PUMI_EXAMPLES_SRCS})
+  get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
+  string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
+  string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
+  set(TEST_NAME ${PFX}${TEST_NAME})

-    set(THIS_TEST_OPTIONS "-no-vis")
-    list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
-    # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")
+  set(THIS_TEST_OPTIONS "-no-vis")
+  list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
+  # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")

-    # All PUMI examples require MPI
-    if (FALSE)
-      add_test(NAME ${TEST_NAME}_ser
-        COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
-    else()
-      set(TEST_NP ${${UP_TEST_NAME}_TEST_NP})
-      add_test(NAME ${TEST_NAME}_np=${TEST_NP}
-        COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${TEST_NP}
-        ${MPIEXEC_PREFLAGS}
-        $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
-        ${MPIEXEC_POSTFLAGS})
-    endif()
-  endforeach()
-endif()
+  # All PUMI examples require MPI
+  if (FALSE)
+    add_test(NAME ${TEST_NAME}_ser
+      COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
+  else()
+    set(TEST_NP ${${UP_TEST_NAME}_TEST_NP})
+    add_test(NAME ${TEST_NAME}_np=${TEST_NP}
+      COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${TEST_NP}
+      ${MPIEXEC_PREFLAGS}
+      $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
+      ${MPIEXEC_POSTFLAGS})
+  endif()
+endforeach()
@@ -145,7 +145,7 @@ int main(int argc, char *argv[])
   // Perform Uniform refinement
   if (ref_levels > 1)
   {
-      auto uniInput = ma::configureUniformRefine(pumi_mesh, ref_levels);
+      ma::Input* uniInput = ma::configureUniformRefine(pumi_mesh, ref_levels);

      if (geom_order > 1)
      {
@@ -150,7 +150,7 @@ int main(int argc, char *argv[])

   if (ref_levels > 1)
   {
-      auto uniInput = ma::configureUniformRefine(pumi_mesh, ref_levels);
+      ma::Input* uniInput = ma::configureUniformRefine(pumi_mesh, ref_levels);

      if ( geom_order > 1)
      {
@@ -345,7 +345,9 @@ int main(int argc, char *argv[])
      apf::destroyField(ipfield);

      // 18. Perform MesAdapt.
-      auto erinput = ma::configure(pumi_mesh, sizefield);
+      ma::Input* erinput = ma::configure(pumi_mesh, sizefield);
+      erinput->shouldFixShape = true;
+      erinput->maximumIterations = 2;
      if ( geom_order > 1)
      {
         crv::adapt(erinput);
@@ -41,38 +41,36 @@ add_mfem_examples(SUNDIALS_EXAMPLES_SRCS ${PFX} "" test_sundials)
 # which builds the examples and runs:
 #   ctest -R sundials

-if (MFEM_ENABLE_TESTING)
-  # Command line options for the tests.
-  # Example 9: test CVODE with CV_ADAMS (non-stiff implicit) time stepping
-  set(EX9_COMMON_OPTS -m ../../data/periodic-hexagon.mesh -p 0 -s 7)
-  set(EX9_TEST_OPTS ${EX9_COMMON_OPTS}  -r  2 -dt 0.0018 -vs 25)
-  set(EX9P_TEST_OPTS ${EX9_COMMON_OPTS} -rp 1 -dt 0.0009 -vs 50)
-  # Example 10: test CVODE with CV_BDF (stiff implicit) time stepping
-  set(EX10_COMMON_OPTS -m ../../data/beam-quad.mesh -o 2 -s 5 -dt 0.15 -tf 6 -vs 10)
-  set(EX10_TEST_OPTS ${EX10_COMMON_OPTS}  -r  2)
-  set(EX10P_TEST_OPTS ${EX10_COMMON_OPTS} -rp 1)
-  # Example 16: use the default options
+# Command line options for the tests.
+# Example 9: test CVODE with CV_ADAMS (non-stiff implicit) time stepping
+set(EX9_COMMON_OPTS -m ../../data/periodic-hexagon.mesh -p 0 -s 7)
+set(EX9_TEST_OPTS ${EX9_COMMON_OPTS}  -r  2 -dt 0.0018 -vs 25)
+set(EX9P_TEST_OPTS ${EX9_COMMON_OPTS} -rp 1 -dt 0.0009 -vs 50)
+# Example 10: test CVODE with CV_BDF (stiff implicit) time stepping
+set(EX10_COMMON_OPTS -m ../../data/beam-quad.mesh -o 2 -s 5 -dt 0.15 -tf 6 -vs 10)
+set(EX10_TEST_OPTS ${EX10_COMMON_OPTS}  -r  2)
+set(EX10P_TEST_OPTS ${EX10_COMMON_OPTS} -rp 1)
+# Example 16: use the default options

-  # Add the tests: one test per source file.
-  foreach(SRC_FILE ${SUNDIALS_EXAMPLES_SRCS})
-    get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
-    string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
-    string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
-    set(TEST_NAME ${PFX}${TEST_NAME})
+# Add the tests: one test per source file.
+foreach(SRC_FILE ${SUNDIALS_EXAMPLES_SRCS})
+  get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
+  string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
+  string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
+  set(TEST_NAME ${PFX}${TEST_NAME})

-    set(THIS_TEST_OPTIONS "-no-vis")
-    list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
-    # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")
+  set(THIS_TEST_OPTIONS "-no-vis")
+  list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
+  # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")

-    if (NOT (${TEST_NAME} MATCHES ".*p$"))
-      add_test(NAME ${TEST_NAME}_ser
-        COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
-    else()
-      add_test(NAME ${TEST_NAME}_np=4
-        COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
-        ${MPIEXEC_PREFLAGS}
-        $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
-        ${MPIEXEC_POSTFLAGS})
-    endif()
-  endforeach()
-endif()
+  if (NOT (${TEST_NAME} MATCHES ".*p$"))
+    add_test(NAME ${TEST_NAME}_ser
+      COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
+  else()
+    add_test(NAME ${TEST_NAME}_np=4
+      COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
+      ${MPIEXEC_PREFLAGS}
+      $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
+      ${MPIEXEC_POSTFLAGS})
+  endif()
+endforeach()
@@ -32,32 +32,31 @@ add_mfem_examples(SUPERLU_EXAMPLES_SRCS ${PFX} "" test_superlu)
 # The SuperLU tests can be run separately using the target "test_superlu"
 # which builds the examples and runs:
 #   ctest -R superlu
-if (MFEM_ENABLE_TESTING)
-  # Command line options for the tests.
-  # Example 1: Test SuperLU on the simple Poisson problem
-  set(EX1_COMMON_OPTS -m ../../data/star.mesh -p 2)
-  set(EX1P_TEST_OPTS ${EX1_COMMON_OPTS})

-  # Add the tests: one test per source file.
-  foreach(SRC_FILE ${SUPERLU_EXAMPLES_SRCS})
-    get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
-    string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
-    string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
-    set(TEST_NAME ${PFX}${TEST_NAME})
+# Command line options for the tests.
+# Example 1: Test SuperLU on the simple Poisson problem
+set(EX1_COMMON_OPTS -m ../../data/star.mesh -p 2)
+set(EX1P_TEST_OPTS ${EX1_COMMON_OPTS})

-    set(THIS_TEST_OPTIONS "-no-vis")
-    list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
-    # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")
+# Add the tests: one test per source file.
+foreach(SRC_FILE ${SUPERLU_EXAMPLES_SRCS})
+  get_filename_component(SRC_FILENAME ${SRC_FILE} NAME)
+  string(REPLACE ".cpp" "" TEST_NAME ${SRC_FILENAME})
+  string(TOUPPER ${TEST_NAME} UP_TEST_NAME)
+  set(TEST_NAME ${PFX}${TEST_NAME})

-    if (NOT (${TEST_NAME} MATCHES ".*p$"))
-      add_test(NAME ${TEST_NAME}_ser
-        COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
-    else()
-      add_test(NAME ${TEST_NAME}_np=4
-        COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
-        ${MPIEXEC_PREFLAGS}
-        $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
-        ${MPIEXEC_POSTFLAGS})
-    endif()
-  endforeach()
-endif()
+  set(THIS_TEST_OPTIONS "-no-vis")
+  list(APPEND THIS_TEST_OPTIONS ${${UP_TEST_NAME}_TEST_OPTS})
+  # message(STATUS "Test ${TEST_NAME} options: ${THIS_TEST_OPTIONS}")
+
+  if (NOT (${TEST_NAME} MATCHES ".*p$"))
+    add_test(NAME ${TEST_NAME}_ser
+      COMMAND ${TEST_NAME} ${THIS_TEST_OPTIONS})
+  else()
+    add_test(NAME ${TEST_NAME}_np=4
+      COMMAND ${MPIEXEC} ${MPIEXEC_NUMPROC_FLAG} ${MFEM_MPI_NP}
+      ${MPIEXEC_PREFLAGS}
+      $<TARGET_FILE:${TEST_NAME}> ${THIS_TEST_OPTIONS}
+      ${MPIEXEC_POSTFLAGS})
+  endif()
+endforeach()
@@ -43,15 +43,6 @@ set(SRCS
  eltrans.cpp
  estimators.cpp
  fe.cpp
-  fe/fe_base.cpp
-  fe/fe_fixed_order.cpp
-  fe/fe_h1.cpp
-  fe/fe_l2.cpp
-  fe/fe_nd.cpp
-  fe/fe_nurbs.cpp
-  fe/fe_pos.cpp
-  fe/fe_rt.cpp
-  fe/fe_ser.cpp
  fe_coll.cpp
  fespace.cpp
  geom.cpp
@@ -133,15 +124,6 @@ set(HDRS
  eltrans.hpp
  estimators.hpp
  fe.hpp
-  fe/fe_base.hpp
-  fe/fe_fixed_order.hpp
-  fe/fe_h1.hpp
-  fe/fe_l2.hpp
-  fe/fe_nd.hpp
-  fe/fe_nurbs.hpp
-  fe/fe_pos.hpp
-  fe/fe_rt.hpp
-  fe/fe_ser.hpp
  fe_coll.hpp
  fem.hpp
  fespace.hpp
@@ -969,7 +969,6 @@ void BilinearForm::EliminateVDofs(const Array<int> &vdofs,
                                  const Vector &sol, Vector &rhs,
                                  DiagonalPolicy dpolicy)
 {
-   vdofs.HostRead();
   for (int i = 0; i < vdofs.Size(); i++)
   {
      int vdof = vdofs[i];
@@ -514,10 +514,6 @@ void EABilinearFormExtension::Assemble()

   Array<BilinearFormIntegrator*> &integrators = *a->GetDBFI();
   const int integratorCount = integrators.Size();
-   if ( integratorCount == 0 )
-   {
-      ea_data = 0.0;
-   }
   for (int i = 0; i < integratorCount; ++i)
   {
      integrators[i]->AssembleEA(*a->FESpace(), ea_data, i);
@@ -747,7 +747,6 @@ void DiffusionIntegrator::AssembleElementMatrix
 #ifdef MFEM_THREAD_SAFE
   DenseMatrix dshape(nd, dim), dshapedxt(nd, spaceDim);
   DenseMatrix dshapedxt_m(nd, MQ ? spaceDim : 0);
-   DenseMatrix M(MQ ? spaceDim : 0);
   Vector D(VQ ? VQ->GetVDim() : 0);
 #else
   dshape.SetSize(nd, dim);
@@ -985,8 +984,6 @@ void DiffusionIntegrator::ComputeElementFlux
                  "Unexpected height for MatrixCoefficient");
   }

-   MFEM_VERIFY(!SMQ, "SymmetricMatrixCoefficient not supported here");
-
 #ifdef MFEM_THREAD_SAFE
   DenseMatrix dshape(nd,dim), invdfdx(dim, spaceDim);
   DenseMatrix M(MQ ? spaceDim : 0);
@@ -999,7 +996,7 @@ void DiffusionIntegrator::ComputeElementFlux
 #endif
   vec.SetSize(dim);
   vecdxt.SetSize(spaceDim);
-   pointflux.SetSize(MQ || VQ ? spaceDim : 0);
+   pointflux.SetSize(MQ ? spaceDim : 0);

   const IntegrationRule &ir = fluxelem.GetNodes();
   fnd = ir.GetNPoints();
@@ -1015,45 +1012,36 @@ void DiffusionIntegrator::ComputeElementFlux
      CalcInverse(Trans.Jacobian(), invdfdx);
      invdfdx.MultTranspose(vec, vecdxt);

-      if (with_coef)
+      if (!MQ && !VQ)
      {
-         if (!MQ && !VQ)
+         if (Q && with_coef)
         {
-            if (Q)
-            {
-               vecdxt *= Q->Eval(Trans,ip);
-            }
-            for (j = 0; j < spaceDim; j++)
-            {
-               flux(fnd*j+i) = vecdxt(j);
-            }
+            vecdxt *= Q->Eval(Trans,ip);
         }
-         else
+         for (j = 0; j < spaceDim; j++)
         {
-            if (MQ)
-            {
-               MQ->Eval(M, Trans, ip);
-               M.Mult(vecdxt, pointflux);
-            }
-            else
-            {
-               VQ->Eval(D, Trans, ip);
-               for (int j=0; j<spaceDim; ++j)
-               {
-                  pointflux[j] = D[j] * vecdxt[j];
-               }
-            }
-            for (j = 0; j < spaceDim; j++)
-            {
-               flux(fnd*j+i) = pointflux(j);
-            }
+            flux(fnd*j+i) = vecdxt(j);
         }
      }
      else
      {
+         if (MQ)
+         {
+            MQ->Eval(M, Trans, ip);
+            M.Mult(vecdxt, pointflux);
+         }
+         else
+         {
+            VQ->Eval(D, Trans, ip);
+            for (int j=0; j<spaceDim; ++j)
+            {
+               pointflux[j] = D[j] * vecdxt[j];
+            }
+
+         }
         for (j = 0; j < spaceDim; j++)
         {
-            flux(fnd*j+i) = vecdxt(j);
+            flux(fnd*j+i) = pointflux(j);
         }
      }
   }
@@ -1069,13 +1057,8 @@ double DiffusionIntegrator::ComputeFluxEnergy

 #ifdef MFEM_THREAD_SAFE
   DenseMatrix M;
-   Vector D(VQ ? VQ->GetVDim() : 0);
-#else
-   D.SetSize(VQ ? VQ->GetVDim() : 0);
 #endif

-   MFEM_VERIFY(!SMQ, "SymmetricMatrixCoefficient not supported here");
-
   shape.SetSize(nd);
   pointflux.SetSize(spaceDim);
   if (d_energy) { vec.SetSize(spaceDim); }
@@ -1104,23 +1087,17 @@ double DiffusionIntegrator::ComputeFluxEnergy
      Trans.SetIntPoint(&ip);
      double w = Trans.Weight() * ip.weight;

-      if (MQ)
-      {
-         MQ->Eval(M, Trans, ip);
-         energy += w * M.InnerProduct(pointflux, pointflux);
-      }
-      else if (VQ)
-      {
-         VQ->Eval(D, Trans, ip);
-         D *= pointflux;
-         energy += w * (D * pointflux);
-      }
-      else
+      if (!MQ)
      {
         double e = (pointflux * pointflux);
         if (Q) { e *= Q->Eval(Trans, ip); }
         energy += w * e;
      }
+      else
+      {
+         MQ->Eval(M, Trans, ip);
+         energy += w * M.InnerProduct(pointflux, pointflux);
+      }

      if (d_energy)
      {
@@ -1130,7 +1107,7 @@ double DiffusionIntegrator::ComputeFluxEnergy
         {
            (*d_energy)[k] += w * vec[k] * vec[k];
         }
-         // TODO: Q, VQ, MQ
+         // TODO: Q, MQ
      }
   }

@@ -1982,32 +1982,24 @@ private:

 public:
   /// Construct a diffusion integrator with coefficient Q = 1
-   DiffusionIntegrator(const IntegrationRule *ir = nullptr)
-      : BilinearFormIntegrator(ir),
-        Q(NULL), VQ(NULL), MQ(NULL), SMQ(NULL), maps(NULL), geom(NULL) { }
+   DiffusionIntegrator()
+      : Q(NULL), VQ(NULL), MQ(NULL), SMQ(NULL), maps(NULL), geom(NULL) { }

   /// Construct a diffusion integrator with a scalar coefficient q
-   DiffusionIntegrator(Coefficient &q, const IntegrationRule *ir = nullptr)
-      : BilinearFormIntegrator(ir),
-        Q(&q), VQ(NULL), MQ(NULL), SMQ(NULL), maps(NULL), geom(NULL) { }
+   DiffusionIntegrator(Coefficient &q)
+      : Q(&q), VQ(NULL), MQ(NULL), SMQ(NULL), maps(NULL), geom(NULL) { }

   /// Construct a diffusion integrator with a vector coefficient q
-   DiffusionIntegrator(VectorCoefficient &q,
-                       const IntegrationRule *ir = nullptr)
-      : BilinearFormIntegrator(ir),
-        Q(NULL), VQ(&q), MQ(NULL), SMQ(NULL), maps(NULL), geom(NULL) { }
+   DiffusionIntegrator(VectorCoefficient &q)
+      : Q(NULL), VQ(&q), MQ(NULL), SMQ(NULL), maps(NULL), geom(NULL) { }

   /// Construct a diffusion integrator with a matrix coefficient q
-   DiffusionIntegrator(MatrixCoefficient &q,
-                       const IntegrationRule *ir = nullptr)
-      : BilinearFormIntegrator(ir),
-        Q(NULL), VQ(NULL), MQ(&q), SMQ(NULL), maps(NULL), geom(NULL) { }
+   DiffusionIntegrator(MatrixCoefficient &q)
+      : Q(NULL), VQ(NULL), MQ(&q), SMQ(NULL), maps(NULL), geom(NULL) { }

   /// Construct a diffusion integrator with a symmetric matrix coefficient q
-   DiffusionIntegrator(SymmetricMatrixCoefficient &q,
-                       const IntegrationRule *ir = nullptr)
-      : BilinearFormIntegrator(ir),
-        Q(NULL), VQ(NULL), MQ(NULL), SMQ(&q), maps(NULL), geom(NULL) { }
+   DiffusionIntegrator(SymmetricMatrixCoefficient &q)
+      : Q(NULL), VQ(NULL), MQ(NULL), SMQ(&q), maps(NULL), geom(NULL) { }

   /** Given a particular Finite Element computes the element stiffness matrix
       elmat. */
@@ -2676,9 +2668,6 @@ public:
   VectorDiffusionIntegrator(Coefficient &q)
      : Q(&q) { }

-   VectorDiffusionIntegrator(Coefficient &q, const IntegrationRule *ir)
-      : BilinearFormIntegrator(ir), Q(&q) { }
-
   /** \brief Integrator with scalar coefficient for caller-specified vector
       dimension.

@@ -2938,11 +2927,10 @@ public:

    sum_e eta (r_e([u]), r_e([v]))

-    where r_e is the lifting operator defined on each edge e (potentially
-    weighted by a coefficient Q). The parameter eta can be chosen to be one to
-    obtain a stable discretization. The constructor for this integrator requires
-    the finite element space because the lifting operator depends on the
-    element-wise inverse mass matrix.
+    where r_e is the lifting operator defined on each edge e. The parameter eta
+    can be chosen to be one to obtain a stable discretization. The constructor
+    for this integrator requires the finite element space because the lifting
+    operator depends on the element-wise inverse mass matrix.

    BR2 stands for the second method of Bassi and Rebay:

@@ -2965,28 +2953,14 @@ protected:
   Array<int> ipiv;
   Array<int> ipiv_offsets, Minv_offsets;

-   Coefficient *Q;
-
   Vector shape1, shape2;

   DenseMatrix R11, R12, R21, R22;
   DenseMatrix MinvR11, MinvR12, MinvR21, MinvR22;
   DenseMatrix Re, MinvRe;

-   /// Precomputes the inverses (LU factorizations) of the local mass matrices.
-   /** @a fes must be a DG space, so the mass matrix is block diagonal, and its
-       inverse can be computed locally. This is required for the computation of
-       the lifting operators @a r_e.
-   */
-   void PrecomputeMassInverse(class FiniteElementSpace &fes);
-
 public:
-   DGDiffusionBR2Integrator(class FiniteElementSpace &fes, double e = 1.0);
-   DGDiffusionBR2Integrator(class FiniteElementSpace &fes, Coefficient &Q_,
-                            double e = 1.0);
-   MFEM_DEPRECATED DGDiffusionBR2Integrator(class FiniteElementSpace *fes,
-                                            double e = 1.0);
-
+   DGDiffusionBR2Integrator(class FiniteElementSpace *fes, double e = 1.0);
   using BilinearFormIntegrator::AssembleFaceMatrix;
   virtual void AssembleFaceMatrix(const FiniteElement &el1,
                                   const FiniteElement &el2,
@@ -16,39 +16,20 @@
 namespace mfem
 {

-DGDiffusionBR2Integrator::DGDiffusionBR2Integrator(
-   FiniteElementSpace &fes, double e) : eta(e), Q(NULL)
+DGDiffusionBR2Integrator::DGDiffusionBR2Integrator(FiniteElementSpace *fes,
+                                                   double e) : eta(e)
 {
-   PrecomputeMassInverse(fes);
-}
-
-DGDiffusionBR2Integrator::DGDiffusionBR2Integrator(
-   FiniteElementSpace &fes, Coefficient &Q_, double e) : eta(e), Q(&Q_)
-{
-   PrecomputeMassInverse(fes);
-}
-
-DGDiffusionBR2Integrator::DGDiffusionBR2Integrator(
-   FiniteElementSpace *fes, double e) : eta(e), Q(NULL)
-{
-   PrecomputeMassInverse(*fes);
-}
-
-void DGDiffusionBR2Integrator::PrecomputeMassInverse(FiniteElementSpace &fes)
-{
-   MFEM_VERIFY(fes.IsDGSpace(),
-               "The BR2 integrator is only defined for DG spaces.");
   // Precompute local mass matrix inverses needed for the lifting operators
   // First compute offsets and total size needed (e.g. for mixed meshes or
   // p-refinement)
-   int nel = fes.GetNE();
+   int nel = fes->GetNE();
   Minv_offsets.SetSize(nel+1);
   ipiv_offsets.SetSize(nel+1);
   ipiv_offsets[0] = 0;
   Minv_offsets[0] = 0;
   for (int i=0; i<nel; ++i)
   {
-      int dof = fes.GetFE(i)->GetDof();
+      int dof = fes->GetFE(i)->GetDof();
      ipiv_offsets[i+1] = ipiv_offsets[i] + dof;
      Minv_offsets[i+1] = Minv_offsets[i] + dof*dof;
   }
@@ -56,7 +37,7 @@ void DGDiffusionBR2Integrator::PrecomputeMassInverse(FiniteElementSpace &fes)
 #ifdef MFEM_USE_MPI
   // When running in parallel, we also need to compute the local mass matrices
   // of face neighbor elements
-   ParFiniteElementSpace *pfes = dynamic_cast<ParFiniteElementSpace *>(&fes);
+   ParFiniteElementSpace *pfes = dynamic_cast<ParFiniteElementSpace *>(fes);
   if (pfes != NULL)
   {
      ParMesh *pmesh = pfes->GetParMesh();
@@ -83,15 +64,15 @@ void DGDiffusionBR2Integrator::PrecomputeMassInverse(FiniteElementSpace &fes)
   {
      const FiniteElement *fe = NULL;
      ElementTransformation *tr = NULL;
-      if (i < fes.GetNE())
+      if (i < fes->GetNE())
      {
-         fe = fes.GetFE(i);
-         tr = fes.GetElementTransformation(i);
+         fe = fes->GetFE(i);
+         tr = fes->GetElementTransformation(i);
      }
      else
      {
 #ifdef MFEM_USE_MPI
-         int inbr = i - fes.GetNE();
+         int inbr = i - fes->GetNE();
         fe = pfes->GetFaceNbrFE(inbr);
         tr = pfes->GetParMesh()->GetFaceNbrElementTransformation(inbr);
 #endif
@@ -170,24 +151,21 @@ void DGDiffusionBR2Integrator::AssembleFaceMatrix(
   for (int p = 0; p < ir->GetNPoints(); p++)
   {
      const IntegrationPoint &ip = ir->IntPoint(p);
-      Trans.SetAllIntPoints(&ip);
+      IntegrationPoint eip1, eip2;

-      const IntegrationPoint &eip1 = Trans.Elem1->GetIntPoint();
+      Trans.Loc1.Transform(ip, eip1);
      el1.CalcShape(eip1, shape1);
-      double q = Q ? Q->Eval(*Trans.Elem1, eip1) : 1.0;
      if (ndof2)
      {
-         const IntegrationPoint &eip2 = Trans.Elem2->GetIntPoint();
+         Trans.Loc2.Transform(ip, eip2);
         el2.CalcShape(eip2, shape2);
-         // Set coefficient value q to the average of the values on either side
-         if (Q) { q = 0.5*(q + Q->Eval(*Trans.Elem2, eip2)); }
      }
-      // Take sqrt here because
-      //    eta (r_e([u]), r_e([v])) = (sqrt(eta) r_e([u]), sqrt(eta) r_e([v]))
-      double w = sqrt((factor + 1)*eta*q)*ip.weight*Trans.Face->Weight();
-      // r_e is defined by, (r_e([u]), tau) = <[u], {tau}>, so we pick up a
-      // factor of 0.5 on interior faces from the average term.
-      if (ndof2) { w *= 0.5; }
+
+      double w = factor*sqrt(eta)*ip.weight*Trans.Face->Weight();
+      if (ndof2)
+      {
+         w /= 2;
+      }

      for (int i = 0; i < ndof1; i++)
      {
@@ -17,14 +17,14 @@ namespace mfem
 {

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EAConvectionAssemble1D(const int NE,
-                                   const Array<double> &b,
-                                   const Array<double> &g,
-                                   const Vector &padata,
-                                   Vector &eadata,
-                                   const bool add,
-                                   const int d1d = 0,
-                                   const int q1d = 0)
+void EAConvectionAssemble1D(const int NE,
+                            const Array<double> &b,
+                            const Array<double> &g,
+                            const Vector &padata,
+                            Vector &eadata,
+                            const bool add,
+                            const int d1d = 0,
+                            const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -69,14 +69,14 @@ static void EAConvectionAssemble1D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EAConvectionAssemble2D(const int NE,
-                                   const Array<double> &b,
-                                   const Array<double> &g,
-                                   const Vector &padata,
-                                   Vector &eadata,
-                                   const bool add,
-                                   const int d1d = 0,
-                                   const int q1d = 0)
+void EAConvectionAssemble2D(const int NE,
+                            const Array<double> &b,
+                            const Array<double> &g,
+                            const Vector &padata,
+                            Vector &eadata,
+                            const bool add,
+                            const int d1d = 0,
+                            const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -146,14 +146,14 @@ static void EAConvectionAssemble2D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EAConvectionAssemble3D(const int NE,
-                                   const Array<double> &b,
-                                   const Array<double> &g,
-                                   const Vector &padata,
-                                   Vector &eadata,
-                                   const bool add,
-                                   const int d1d = 0,
-                                   const int q1d = 0)
+void EAConvectionAssemble3D(const int NE,
+                            const Array<double> &b,
+                            const Array<double> &g,
+                            const Vector &padata,
+                            Vector &eadata,
+                            const bool add,
+                            const int d1d = 0,
+                            const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -21,13 +21,13 @@ namespace mfem
 // PA Convection Integrator

 // PA Convection Assemble 2D kernel
-static void PAConvectionSetup2D(const int NQ,
-                                const int NE,
-                                const Array<double> &w,
-                                const Vector &j,
-                                const Vector &vel,
-                                const double alpha,
-                                Vector &op)
+void PAConvectionSetup2D(const int NQ,
+                         const int NE,
+                         const Array<double> &w,
+                         const Vector &j,
+                         const Vector &vel,
+                         const double alpha,
+                         Vector &op)
 {
   constexpr int DIM = 2;

@@ -60,13 +60,13 @@ static void PAConvectionSetup2D(const int NQ,
 }

 // PA Convection Assemble 3D kernel
-static void PAConvectionSetup3D(const int NQ,
-                                const int NE,
-                                const Array<double> &w,
-                                const Vector &j,
-                                const Vector &vel,
-                                const double alpha,
-                                Vector &op)
+void PAConvectionSetup3D(const int NQ,
+                         const int NE,
+                         const Array<double> &w,
+                         const Vector &j,
+                         const Vector &vel,
+                         const double alpha,
+                         Vector &op)
 {
   constexpr int DIM = 3;
   constexpr int SDIM = DIM;
@@ -135,7 +135,7 @@ static void PAConvectionSetup(const int dim,
 }

 // PA Convection Apply 2D kernel
-template<int T_D1D = 0, int T_Q1D = 0> static
+template<int T_D1D = 0, int T_Q1D = 0>
 void PAConvectionApply2D(const int ne,
                         const Array<double> &b,
                         const Array<double> &g,
@@ -254,7 +254,7 @@ void PAConvectionApply2D(const int ne,
 }

 // Optimized PA Convection Apply 2D kernel
-template<int T_D1D = 0, int T_Q1D = 0, int T_NBZ = 0> static
+template<int T_D1D = 0, int T_Q1D = 0, int T_NBZ = 0>
 void SmemPAConvectionApply2D(const int ne,
                             const Array<double> &b,
                             const Array<double> &g,
@@ -382,7 +382,7 @@ void SmemPAConvectionApply2D(const int ne,
 }

 // PA Convection Apply 3D kernel
-template<int T_D1D = 0, int T_Q1D = 0> static
+template<int T_D1D = 0, int T_Q1D = 0>
 void PAConvectionApply3D(const int ne,
                         const Array<double> &b,
                         const Array<double> &g,
@@ -563,7 +563,7 @@ void PAConvectionApply3D(const int ne,
 }

 // Optimized PA Convection Apply 3D kernel
-template<int T_D1D = 0, int T_Q1D = 0> static
+template<int T_D1D = 0, int T_Q1D = 0>
 void SmemPAConvectionApply3D(const int ne,
                             const Array<double> &b,
                             const Array<double> &g,
@@ -16,12 +16,12 @@
 namespace mfem
 {

-static void EADGTraceAssemble1DInt(const int NF,
-                                   const Array<double> &basis,
-                                   const Vector &padata,
-                                   Vector &eadata_int,
-                                   Vector &eadata_ext,
-                                   const bool add)
+void EADGTraceAssemble1DInt(const int NF,
+                            const Array<double> &basis,
+                            const Vector &padata,
+                            Vector &eadata_int,
+                            Vector &eadata_ext,
+                            const bool add)
 {
   auto D = Reshape(padata.Read(), 2, 2, NF);
   auto A_int = Reshape(eadata_int.ReadWrite(), 2, NF);
@@ -50,11 +50,11 @@ static void EADGTraceAssemble1DInt(const int NF,
   });
 }

-static void EADGTraceAssemble1DBdr(const int NF,
-                                   const Array<double> &basis,
-                                   const Vector &padata,
-                                   Vector &eadata_bdr,
-                                   const bool add)
+void EADGTraceAssemble1DBdr(const int NF,
+                            const Array<double> &basis,
+                            const Vector &padata,
+                            Vector &eadata_bdr,
+                            const bool add)
 {
   auto D = Reshape(padata.Read(), 2, 2, NF);
   auto A_bdr = Reshape(eadata_bdr.ReadWrite(), NF);
@@ -72,14 +72,14 @@ static void EADGTraceAssemble1DBdr(const int NF,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EADGTraceAssemble2DInt(const int NF,
-                                   const Array<double> &basis,
-                                   const Vector &padata,
-                                   Vector &eadata_int,
-                                   Vector &eadata_ext,
-                                   const bool add,
-                                   const int d1d = 0,
-                                   const int q1d = 0)
+void EADGTraceAssemble2DInt(const int NF,
+                            const Array<double> &basis,
+                            const Vector &padata,
+                            Vector &eadata_int,
+                            Vector &eadata_ext,
+                            const bool add,
+                            const int d1d = 0,
+                            const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -128,13 +128,13 @@ static void EADGTraceAssemble2DInt(const int NF,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EADGTraceAssemble2DBdr(const int NF,
-                                   const Array<double> &basis,
-                                   const Vector &padata,
-                                   Vector &eadata_bdr,
-                                   const bool add,
-                                   const int d1d = 0,
-                                   const int q1d = 0)
+void EADGTraceAssemble2DBdr(const int NF,
+                            const Array<double> &basis,
+                            const Vector &padata,
+                            Vector &eadata_bdr,
+                            const bool add,
+                            const int d1d = 0,
+                            const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -170,14 +170,14 @@ static void EADGTraceAssemble2DBdr(const int NF,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EADGTraceAssemble3DInt(const int NF,
-                                   const Array<double> &basis,
-                                   const Vector &padata,
-                                   Vector &eadata_int,
-                                   Vector &eadata_ext,
-                                   const bool add,
-                                   const int d1d = 0,
-                                   const int q1d = 0)
+void EADGTraceAssemble3DInt(const int NF,
+                            const Array<double> &basis,
+                            const Vector &padata,
+                            Vector &eadata_int,
+                            Vector &eadata_ext,
+                            const bool add,
+                            const int d1d = 0,
+                            const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -268,13 +268,13 @@ static void EADGTraceAssemble3DInt(const int NF,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EADGTraceAssemble3DBdr(const int NF,
-                                   const Array<double> &basis,
-                                   const Vector &padata,
-                                   Vector &eadata_bdr,
-                                   const bool add,
-                                   const int d1d = 0,
-                                   const int q1d = 0)
+void EADGTraceAssemble3DBdr(const int NF,
+                            const Array<double> &basis,
+                            const Vector &padata,
+                            Vector &eadata_bdr,
+                            const bool add,
+                            const int d1d = 0,
+                            const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -19,16 +19,16 @@ using namespace std;
 namespace mfem
 {
 // PA DG Trace Integrator
-static void PADGTraceSetup2D(const int Q1D,
-                             const int NF,
-                             const Array<double> &w,
-                             const Vector &det,
-                             const Vector &nor,
-                             const Vector &rho,
-                             const Vector &vel,
-                             const double alpha,
-                             const double beta,
-                             Vector &op)
+void PADGTraceSetup2D(const int Q1D,
+                      const int NF,
+                      const Array<double> &w,
+                      const Vector &det,
+                      const Vector &nor,
+                      const Vector &rho,
+                      const Vector &vel,
+                      const double alpha,
+                      const double beta,
+                      Vector &op)
 {
   const int VDIM = 2;

@@ -61,16 +61,16 @@ static void PADGTraceSetup2D(const int Q1D,
   });
 }

-static void PADGTraceSetup3D(const int Q1D,
-                             const int NF,
-                             const Array<double> &w,
-                             const Vector &det,
-                             const Vector &nor,
-                             const Vector &rho,
-                             const Vector &vel,
-                             const double alpha,
-                             const double beta,
-                             Vector &op)
+void PADGTraceSetup3D(const int Q1D,
+                      const int NF,
+                      const Array<double> &w,
+                      const Vector &det,
+                      const Vector &nor,
+                      const Vector &rho,
+                      const Vector &vel,
+                      const double alpha,
+                      const double beta,
+                      Vector &op)
 {
   const int VDIM = 3;

@@ -301,7 +301,7 @@ void DGTraceIntegrator::AssemblePABoundaryFaces(const FiniteElementSpace& fes)
 }

 // PA DGTrace Apply 2D kernel for Gauss-Lobatto/Bernstein
-template<int T_D1D = 0, int T_Q1D = 0> static
+template<int T_D1D = 0, int T_Q1D = 0>
 void PADGTraceApply2D(const int NF,
                      const Array<double> &b,
                      const Array<double> &bt,
@@ -392,7 +392,7 @@ void PADGTraceApply2D(const int NF,
 }

 // PA DGTrace Apply 3D kernel for Gauss-Lobatto/Bernstein
-template<int T_D1D = 0, int T_Q1D = 0> static
+template<int T_D1D = 0, int T_Q1D = 0>
 void PADGTraceApply3D(const int NF,
                      const Array<double> &b,
                      const Array<double> &bt,
@@ -537,7 +537,7 @@ void PADGTraceApply3D(const int NF,
 }

 // Optimized PA DGTrace Apply 3D kernel for Gauss-Lobatto/Bernstein
-template<int T_D1D = 0, int T_Q1D = 0, int T_NBZ = 0> static
+template<int T_D1D = 0, int T_Q1D = 0, int T_NBZ = 0>
 void SmemPADGTraceApply3D(const int NF,
                          const Array<double> &b,
                          const Array<double> &bt,
@@ -701,7 +701,7 @@ static void PADGTraceApply(const int dim,
 }

 // PA DGTrace Apply 2D kernel for Gauss-Lobatto/Bernstein
-template<int T_D1D = 0, int T_Q1D = 0> static
+template<int T_D1D = 0, int T_Q1D = 0>
 void PADGTraceApplyTranspose2D(const int NF,
                               const Array<double> &b,
                               const Array<double> &bt,
@@ -797,7 +797,7 @@ void PADGTraceApplyTranspose2D(const int NF,
 }

 // PA DGTrace Apply Transpose 3D kernel for Gauss-Lobatto/Bernstein
-template<int T_D1D = 0, int T_Q1D = 0> static
+template<int T_D1D = 0, int T_Q1D = 0>
 void PADGTraceApplyTranspose3D(const int NF,
                               const Array<double> &b,
                               const Array<double> &bt,
@@ -953,7 +953,7 @@ void PADGTraceApplyTranspose3D(const int NF,
 }

 // Optimized PA DGTrace Apply Transpose 3D kernel for Gauss-Lobatto/Bernstein
-template<int T_D1D = 0, int T_Q1D = 0, int T_NBZ = 0> static
+template<int T_D1D = 0, int T_Q1D = 0, int T_NBZ = 0>
 void SmemPADGTraceApplyTranspose3D(const int NF,
                                   const Array<double> &b,
                                   const Array<double> &bt,
@@ -17,14 +17,14 @@ namespace mfem
 {

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EADiffusionAssemble1D(const int NE,
-                                  const Array<double> &b,
-                                  const Array<double> &g,
-                                  const Vector &padata,
-                                  Vector &eadata,
-                                  const bool add,
-                                  const int d1d = 0,
-                                  const int q1d = 0)
+void EADiffusionAssemble1D(const int NE,
+                           const Array<double> &b,
+                           const Array<double> &g,
+                           const Vector &padata,
+                           Vector &eadata,
+                           const bool add,
+                           const int d1d = 0,
+                           const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -68,14 +68,14 @@ static void EADiffusionAssemble1D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EADiffusionAssemble2D(const int NE,
-                                  const Array<double> &b,
-                                  const Array<double> &g,
-                                  const Vector &padata,
-                                  Vector &eadata,
-                                  const bool add,
-                                  const int d1d = 0,
-                                  const int q1d = 0)
+void EADiffusionAssemble2D(const int NE,
+                           const Array<double> &b,
+                           const Array<double> &g,
+                           const Vector &padata,
+                           Vector &eadata,
+                           const bool add,
+                           const int d1d = 0,
+                           const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -145,14 +145,14 @@ static void EADiffusionAssemble2D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EADiffusionAssemble3D(const int NE,
-                                  const Array<double> &b,
-                                  const Array<double> &g,
-                                  const Vector &padata,
-                                  Vector &eadata,
-                                  const bool add,
-                                  const int d1d = 0,
-                                  const int q1d = 0)
+void EADiffusionAssemble3D(const int NE,
+                           const Array<double> &b,
+                           const Array<double> &g,
+                           const Vector &padata,
+                           Vector &eadata,
+                           const bool add,
+                           const int d1d = 0,
+                           const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -125,7 +125,7 @@ void PADiffusionSetup2D<2>(const int Q1D,
               D(qx,qy,0,e) = w_detJ * ( J22*R11 - J12*R21); // 1,1
               D(qx,qy,1,e) = w_detJ * (-J21*R11 + J11*R21); // 2,1
               D(qx,qy,2,e) = w_detJ * (symmetric ? (-J21*R12 + J11*R22) :
-                                        (J22*R12 - J12*R22)); // 2,2 or 1,2
+               (J22*R12 - J12*R22)); // 2,2 or 1,2
               if (!symmetric)
               {
                  D(qx,qy,3,e) = w_detJ * (-J21*R12 + J11*R22); // 2,2
@@ -496,14 +496,14 @@ void DiffusionIntegrator::AssemblePA(const FiniteElementSpace &fes)
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void PADiffusionDiagonal2D(const int NE,
-                                  const bool symmetric,
-                                  const Array<double> &b,
-                                  const Array<double> &g,
-                                  const Vector &d,
-                                  Vector &y,
-                                  const int d1d = 0,
-                                  const int q1d = 0)
+void PADiffusionDiagonal2D(const int NE,
+                           const bool symmetric,
+                           const Array<double> &b,
+                           const Array<double> &g,
+                           const Vector &d,
+                           Vector &y,
+                           const int d1d = 0,
+                           const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -562,14 +562,14 @@ static void PADiffusionDiagonal2D(const int NE,

 // Shared memory PA Diffusion Diagonal 2D kernel
 template<int T_D1D = 0, int T_Q1D = 0, int T_NBZ = 0>
-static void SmemPADiffusionDiagonal2D(const int NE,
-                                      const bool symmetric,
-                                      const Array<double> &b_,
-                                      const Array<double> &g_,
-                                      const Vector &d_,
-                                      Vector &y_,
-                                      const int d1d = 0,
-                                      const int q1d = 0)
+void SmemPADiffusionDiagonal2D(const int NE,
+                               const bool symmetric,
+                               const Array<double> &b_,
+                               const Array<double> &g_,
+                               const Vector &d_,
+                               Vector &y_,
+                               const int d1d = 0,
+                               const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -656,14 +656,14 @@ static void SmemPADiffusionDiagonal2D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void PADiffusionDiagonal3D(const int NE,
-                                  const bool symmetric,
-                                  const Array<double> &b,
-                                  const Array<double> &g,
-                                  const Vector &d,
-                                  Vector &y,
-                                  const int d1d = 0,
-                                  const int q1d = 0)
+void PADiffusionDiagonal3D(const int NE,
+                           const bool symmetric,
+                           const Array<double> &b,
+                           const Array<double> &g,
+                           const Vector &d,
+                           Vector &y,
+                           const int d1d = 0,
+                           const int q1d = 0)
 {
   constexpr int DIM = 3;
   const int D1D = T_D1D ? T_D1D : d1d;
@@ -757,14 +757,14 @@ static void PADiffusionDiagonal3D(const int NE,

 // Shared memory PA Diffusion Diagonal 3D kernel
 template<int T_D1D = 0, int T_Q1D = 0>
-static void SmemPADiffusionDiagonal3D(const int NE,
-                                      const bool symmetric,
-                                      const Array<double> &b_,
-                                      const Array<double> &g_,
-                                      const Vector &d_,
-                                      Vector &y_,
-                                      const int d1d = 0,
-                                      const int q1d = 0)
+void SmemPADiffusionDiagonal3D(const int NE,
+                               const bool symmetric,
+                               const Array<double> &b_,
+                               const Array<double> &g_,
+                               const Vector &d_,
+                               Vector &y_,
+                               const int d1d = 0,
+                               const int q1d = 0)
 {
   constexpr int DIM = 3;
   const int D1D = T_D1D ? T_D1D : d1d;
@@ -903,11 +903,9 @@ static void PADiffusionAssembleDiagonal(const int dim,
   {
      switch ((D1D << 4 ) | Q1D)
      {
-         case 0x22: return SmemPADiffusionDiagonal3D<2,2>(NE,symm,B,G,D,Y);
         case 0x23: return SmemPADiffusionDiagonal3D<2,3>(NE,symm,B,G,D,Y);
         case 0x34: return SmemPADiffusionDiagonal3D<3,4>(NE,symm,B,G,D,Y);
         case 0x45: return SmemPADiffusionDiagonal3D<4,5>(NE,symm,B,G,D,Y);
-         case 0x46: return SmemPADiffusionDiagonal3D<4,6>(NE,symm,B,G,D,Y);
         case 0x56: return SmemPADiffusionDiagonal3D<5,6>(NE,symm,B,G,D,Y);
         case 0x67: return SmemPADiffusionDiagonal3D<6,7>(NE,symm,B,G,D,Y);
         case 0x78: return SmemPADiffusionDiagonal3D<7,8>(NE,symm,B,G,D,Y);
@@ -1036,17 +1034,17 @@ static void OccaPADiffusionApply3D(const int D1D,

 // PA Diffusion Apply 2D kernel
 template<int T_D1D = 0, int T_Q1D = 0>
-static void PADiffusionApply2D(const int NE,
-                               const bool symmetric,
-                               const Array<double> &b_,
-                               const Array<double> &g_,
-                               const Array<double> &bt_,
-                               const Array<double> &gt_,
-                               const Vector &d_,
-                               const Vector &x_,
-                               Vector &y_,
-                               const int d1d = 0,
-                               const int q1d = 0)
+void PADiffusionApply2D(const int NE,
+                        const bool symmetric,
+                        const Array<double> &b_,
+                        const Array<double> &g_,
+                        const Array<double> &bt_,
+                        const Array<double> &gt_,
+                        const Vector &d_,
+                        const Vector &x_,
+                        Vector &y_,
+                        const int d1d = 0,
+                        const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -1158,15 +1156,15 @@ static void PADiffusionApply2D(const int NE,

 // Shared memory PA Diffusion Apply 2D kernel
 template<int T_D1D = 0, int T_Q1D = 0, int T_NBZ = 0>
-static void SmemPADiffusionApply2D(const int NE,
-                                   const bool symmetric,
-                                   const Array<double> &b_,
-                                   const Array<double> &g_,
-                                   const Vector &d_,
-                                   const Vector &x_,
-                                   Vector &y_,
-                                   const int d1d = 0,
-                                   const int q1d = 0)
+void SmemPADiffusionApply2D(const int NE,
+                            const bool symmetric,
+                            const Array<double> &b_,
+                            const Array<double> &g_,
+                            const Vector &d_,
+                            const Vector &x_,
+                            Vector &y_,
+                            const int d1d = 0,
+                            const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -1316,16 +1314,16 @@ static void SmemPADiffusionApply2D(const int NE,

 // PA Diffusion Apply 3D kernel
 template<int T_D1D = 0, int T_Q1D = 0>
-static void PADiffusionApply3D(const int NE,
-                               const bool symmetric,
-                               const Array<double> &b,
-                               const Array<double> &g,
-                               const Array<double> &bt,
-                               const Array<double> &gt,
-                               const Vector &d_,
-                               const Vector &x_,
-                               Vector &y_,
-                               int d1d = 0, int q1d = 0)
+void PADiffusionApply3D(const int NE,
+                        const bool symmetric,
+                        const Array<double> &b,
+                        const Array<double> &g,
+                        const Array<double> &bt,
+                        const Array<double> &gt,
+                        const Vector &d_,
+                        const Vector &x_,
+                        Vector &y_,
+                        int d1d = 0, int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -1535,15 +1533,15 @@ static MFEM_HOST_DEVICE inline double sign(const int q, const int d)
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void SmemPADiffusionApply3D(const int NE,
-                                   const bool symmetric,
-                                   const Array<double> &b_,
-                                   const Array<double> &g_,
-                                   const Vector &d_,
-                                   const Vector &x_,
-                                   Vector &y_,
-                                   const int d1d = 0,
-                                   const int q1d = 0)
+void SmemPADiffusionApply3D(const int NE,
+                            const bool symmetric,
+                            const Array<double> &b_,
+                            const Array<double> &g_,
+                            const Vector &d_,
+                            const Vector &x_,
+                            Vector &y_,
+                            const int d1d = 0,
+                            const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -1556,7 +1554,7 @@ static void SmemPADiffusionApply3D(const int NE,
   auto d = Reshape(d_.Read(), Q1D, Q1D, Q1D, symmetric ? 6 : 9, NE);
   auto x = Reshape(x_.Read(), D1D, D1D, D1D, NE);
   auto y = Reshape(y_.ReadWrite(), D1D, D1D, D1D, NE);
-   MFEM_FORALL_3D(e, NE, Q1D, Q1D, Q1D,
+   MFEM_FORALL_3D(e, NE, Q1D, Q1D, 1,
   {
      const int D1D = T_D1D ? T_D1D : d1d;
      const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -1585,102 +1583,118 @@ static void SmemPADiffusionApply3D(const int NE,
      double (*QDD0)[MD1][MD1] = (double (*)[MD1][MD1]) (sm0+0);
      double (*QDD1)[MD1][MD1] = (double (*)[MD1][MD1]) (sm0+1);
      double (*QDD2)[MD1][MD1] = (double (*)[MD1][MD1]) (sm0+2);
-      MFEM_FOREACH_THREAD(dz,z,D1D)
+      MFEM_FOREACH_THREAD(dy,y,D1D)
      {
-         MFEM_FOREACH_THREAD(dy,y,D1D)
+         MFEM_FOREACH_THREAD(dx,x,D1D)
         {
-            MFEM_FOREACH_THREAD(dx,x,D1D)
+            MFEM_UNROLL(MD1)
+            for (int dz = 0; dz < D1D; ++dz)
            {
               X[dz][dy][dx] = x(dx,dy,dz,e);
            }
         }
-      }
-      if (MFEM_THREAD_ID(z) == 0)
-      {
-         MFEM_FOREACH_THREAD(dy,y,D1D)
+         MFEM_FOREACH_THREAD(qx,x,Q1D)
         {
-            MFEM_FOREACH_THREAD(qx,x,Q1D)
-            {
-               const int i = qi(qx,dy,Q1D);
-               const int j = dj(qx,dy,D1D);
-               const int k = qk(qx,dy,Q1D);
-               const int l = dl(qx,dy,D1D);
-               B[i][j] = b(qx,dy);
-               G[k][l] = g(qx,dy) * sign(qx,dy);
-            }
+            const int i = qi(qx,dy,Q1D);
+            const int j = dj(qx,dy,D1D);
+            const int k = qk(qx,dy,Q1D);
+            const int l = dl(qx,dy,D1D);
+            B[i][j] = b(qx,dy);
+            G[k][l] = g(qx,dy) * sign(qx,dy);
         }
      }
      MFEM_SYNC_THREAD;
-      MFEM_FOREACH_THREAD(dz,z,D1D)
+      MFEM_FOREACH_THREAD(dy,y,D1D)
      {
-         MFEM_FOREACH_THREAD(dy,y,D1D)
+         MFEM_FOREACH_THREAD(qx,x,Q1D)
         {
-            MFEM_FOREACH_THREAD(qx,x,Q1D)
+            double u[D1D], v[D1D];
+            MFEM_UNROLL(MD1)
+            for (int dz = 0; dz < D1D; dz++) { u[dz] = v[dz] = 0.0; }
+            MFEM_UNROLL(MD1)
+            for (int dx = 0; dx < D1D; ++dx)
            {
-               double u = 0.0, v = 0.0;
-               MFEM_UNROLL(MD1)
-               for (int dx = 0; dx < D1D; ++dx)
-               {
-                  const int i = qi(qx,dx,Q1D);
-                  const int j = dj(qx,dx,D1D);
-                  const int k = qk(qx,dx,Q1D);
-                  const int l = dl(qx,dx,D1D);
-                  const double s = sign(qx,dx);
-                  const double coords = X[dz][dy][dx];
-                  u += coords * B[i][j];
-                  v += coords * G[k][l] * s;
-               }
-               DDQ0[dz][dy][qx] = u;
-               DDQ1[dz][dy][qx] = v;
-            }
-         }
-      }
-      MFEM_SYNC_THREAD;
-      MFEM_FOREACH_THREAD(dz,z,D1D)
-      {
-         MFEM_FOREACH_THREAD(qy,y,Q1D)
-         {
-            MFEM_FOREACH_THREAD(qx,x,Q1D)
-            {
-               double u = 0.0, v = 0.0, w = 0.0;
-               MFEM_UNROLL(MD1)
-               for (int dy = 0; dy < D1D; ++dy)
-               {
-                  const int i = qi(qy,dy,Q1D);
-                  const int j = dj(qy,dy,D1D);
-                  const int k = qk(qy,dy,Q1D);
-                  const int l = dl(qy,dy,D1D);
-                  const double s = sign(qy,dy);
-                  u += DDQ1[dz][dy][qx] * B[i][j];
-                  v += DDQ0[dz][dy][qx] * G[k][l] * s;
-                  w += DDQ0[dz][dy][qx] * B[i][j];
-               }
-               DQQ0[dz][qy][qx] = u;
-               DQQ1[dz][qy][qx] = v;
-               DQQ2[dz][qy][qx] = w;
-            }
-         }
-      }
-      MFEM_SYNC_THREAD;
-      MFEM_FOREACH_THREAD(qz,z,Q1D)
-      {
-         MFEM_FOREACH_THREAD(qy,y,Q1D)
-         {
-            MFEM_FOREACH_THREAD(qx,x,Q1D)
-            {
-               double u = 0.0, v = 0.0, w = 0.0;
+               const int i = qi(qx,dx,Q1D);
+               const int j = dj(qx,dx,D1D);
+               const int k = qk(qx,dx,Q1D);
+               const int l = dl(qx,dx,D1D);
+               const double s = sign(qx,dx);
               MFEM_UNROLL(MD1)
               for (int dz = 0; dz < D1D; ++dz)
+               {
+                  const double coords = X[dz][dy][dx];
+                  u[dz] += coords * B[i][j];
+                  v[dz] += coords * G[k][l] * s;
+               }
+            }
+            MFEM_UNROLL(MD1)
+            for (int dz = 0; dz < D1D; ++dz)
+            {
+               DDQ0[dz][dy][qx] = u[dz];
+               DDQ1[dz][dy][qx] = v[dz];
+            }
+         }
+      }
+      MFEM_SYNC_THREAD;
+      MFEM_FOREACH_THREAD(qy,y,Q1D)
+      {
+         MFEM_FOREACH_THREAD(qx,x,Q1D)
+         {
+            double u[D1D], v[D1D], w[D1D];
+            MFEM_UNROLL(MD1)
+            for (int dz = 0; dz < D1D; dz++) { u[dz] = v[dz] = w[dz] = 0.0; }
+            MFEM_UNROLL(MD1)
+            for (int dy = 0; dy < D1D; ++dy)
+            {
+               const int i = qi(qy,dy,Q1D);
+               const int j = dj(qy,dy,D1D);
+               const int k = qk(qy,dy,Q1D);
+               const int l = dl(qy,dy,D1D);
+               const double s = sign(qy,dy);
+               MFEM_UNROLL(MD1)
+               for (int dz = 0; dz < D1D; dz++)
+               {
+                  u[dz] += DDQ1[dz][dy][qx] * B[i][j];
+                  v[dz] += DDQ0[dz][dy][qx] * G[k][l] * s;
+                  w[dz] += DDQ0[dz][dy][qx] * B[i][j];
+               }
+            }
+            MFEM_UNROLL(MD1)
+            for (int dz = 0; dz < D1D; dz++)
+            {
+               DQQ0[dz][qy][qx] = u[dz];
+               DQQ1[dz][qy][qx] = v[dz];
+               DQQ2[dz][qy][qx] = w[dz];
+            }
+         }
+      }
+      MFEM_SYNC_THREAD;
+      MFEM_FOREACH_THREAD(qy,y,Q1D)
+      {
+         MFEM_FOREACH_THREAD(qx,x,Q1D)
+         {
+            double u[Q1D], v[Q1D], w[Q1D];
+            MFEM_UNROLL(MQ1)
+            for (int qz = 0; qz < Q1D; qz++) { u[qz] = v[qz] = w[qz] = 0.0; }
+            MFEM_UNROLL(MD1)
+            for (int dz = 0; dz < D1D; ++dz)
+            {
+               MFEM_UNROLL(MQ1)
+               for (int qz = 0; qz < Q1D; qz++)
               {
                  const int i = qi(qz,dz,Q1D);
                  const int j = dj(qz,dz,D1D);
                  const int k = qk(qz,dz,Q1D);
                  const int l = dl(qz,dz,D1D);
                  const double s = sign(qz,dz);
-                  u += DQQ0[dz][qy][qx] * B[i][j];
-                  v += DQQ1[dz][qy][qx] * B[i][j];
-                  w += DQQ2[dz][qy][qx] * G[k][l] * s;
+                  u[qz] += DQQ0[dz][qy][qx] * B[i][j];
+                  v[qz] += DQQ1[dz][qy][qx] * B[i][j];
+                  w[qz] += DQQ2[dz][qy][qx] * G[k][l] * s;
               }
+            }
+            MFEM_UNROLL(MQ1)
+            for (int qz = 0; qz < Q1D; qz++)
+            {
               const double O11 = d(qx,qy,qz,0,e);
               const double O12 = d(qx,qy,qz,1,e);
               const double O13 = d(qx,qy,qz,2,e);
@@ -1690,9 +1704,9 @@ static void SmemPADiffusionApply3D(const int NE,
               const double O31 = symmetric ? O13 : d(qx,qy,qz,6,e);
               const double O32 = symmetric ? O23 : d(qx,qy,qz,7,e);
               const double O33 = symmetric ? d(qx,qy,qz,5,e) : d(qx,qy,qz,8,e);
-               const double gX = u;
-               const double gY = v;
-               const double gZ = w;
+               const double gX = u[qz];
+               const double gY = v[qz];
+               const double gZ = w[qz];
               QQQ0[qz][qy][qx] = (O11*gX) + (O12*gY) + (O13*gZ);
               QQQ1[qz][qy][qx] = (O21*gX) + (O22*gY) + (O23*gZ);
               QQQ2[qz][qy][qx] = (O31*gX) + (O32*gY) + (O33*gZ);
@@ -1700,94 +1714,112 @@ static void SmemPADiffusionApply3D(const int NE,
         }
      }
      MFEM_SYNC_THREAD;
-      if (MFEM_THREAD_ID(z) == 0)
+      MFEM_FOREACH_THREAD(d,y,D1D)
      {
-         MFEM_FOREACH_THREAD(d,y,D1D)
+         MFEM_FOREACH_THREAD(q,x,Q1D)
         {
-            MFEM_FOREACH_THREAD(q,x,Q1D)
-            {
-               const int i = qi(q,d,Q1D);
-               const int j = dj(q,d,D1D);
-               const int k = qk(q,d,Q1D);
-               const int l = dl(q,d,D1D);
-               Bt[j][i] = b(q,d);
-               Gt[l][k] = g(q,d) * sign(q,d);
-            }
+            const int i = qi(q,d,Q1D);
+            const int j = dj(q,d,D1D);
+            const int k = qk(q,d,Q1D);
+            const int l = dl(q,d,D1D);
+            Bt[j][i] = b(q,d);
+            Gt[l][k] = g(q,d) * sign(q,d);
         }
      }
      MFEM_SYNC_THREAD;
-      MFEM_FOREACH_THREAD(qz,z,Q1D)
+      MFEM_FOREACH_THREAD(qy,y,Q1D)
      {
-         MFEM_FOREACH_THREAD(qy,y,Q1D)
+         MFEM_FOREACH_THREAD(dx,x,D1D)
         {
-            MFEM_FOREACH_THREAD(dx,x,D1D)
+            double u[Q1D], v[Q1D], w[Q1D];
+            MFEM_UNROLL(MQ1)
+            for (int qz = 0; qz < Q1D; ++qz) { u[qz] = v[qz] = w[qz] = 0.0; }
+            MFEM_UNROLL(MQ1)
+            for (int qx = 0; qx < Q1D; ++qx)
            {
-               double u = 0.0, v = 0.0, w = 0.0;
-               MFEM_UNROLL(MQ1)
-               for (int qx = 0; qx < Q1D; ++qx)
-               {
-                  const int i = qi(qx,dx,Q1D);
-                  const int j = dj(qx,dx,D1D);
-                  const int k = qk(qx,dx,Q1D);
-                  const int l = dl(qx,dx,D1D);
-                  const double s = sign(qx,dx);
-                  u += QQQ0[qz][qy][qx] * Gt[l][k] * s;
-                  v += QQQ1[qz][qy][qx] * Bt[j][i];
-                  w += QQQ2[qz][qy][qx] * Bt[j][i];
-               }
-               QQD0[qz][qy][dx] = u;
-               QQD1[qz][qy][dx] = v;
-               QQD2[qz][qy][dx] = w;
-            }
-         }
-      }
-      MFEM_SYNC_THREAD;
-      MFEM_FOREACH_THREAD(qz,z,Q1D)
-      {
-         MFEM_FOREACH_THREAD(dy,y,D1D)
-         {
-            MFEM_FOREACH_THREAD(dx,x,D1D)
-            {
-               double u = 0.0, v = 0.0, w = 0.0;
-               MFEM_UNROLL(Q1D)
-               for (int qy = 0; qy < Q1D; ++qy)
-               {
-                  const int i = qi(qy,dy,Q1D);
-                  const int j = dj(qy,dy,D1D);
-                  const int k = qk(qy,dy,Q1D);
-                  const int l = dl(qy,dy,D1D);
-                  const double s = sign(qy,dy);
-                  u += QQD0[qz][qy][dx] * Bt[j][i];
-                  v += QQD1[qz][qy][dx] * Gt[l][k] * s;
-                  w += QQD2[qz][qy][dx] * Bt[j][i];
-               }
-               QDD0[qz][dy][dx] = u;
-               QDD1[qz][dy][dx] = v;
-               QDD2[qz][dy][dx] = w;
-            }
-         }
-      }
-      MFEM_SYNC_THREAD;
-      MFEM_FOREACH_THREAD(dz,z,D1D)
-      {
-         MFEM_FOREACH_THREAD(dy,y,D1D)
-         {
-            MFEM_FOREACH_THREAD(dx,x,D1D)
-            {
-               double u = 0.0, v = 0.0, w = 0.0;
+               const int i = qi(qx,dx,Q1D);
+               const int j = dj(qx,dx,D1D);
+               const int k = qk(qx,dx,Q1D);
+               const int l = dl(qx,dx,D1D);
+               const double s = sign(qx,dx);
               MFEM_UNROLL(MQ1)
               for (int qz = 0; qz < Q1D; ++qz)
+               {
+                  u[qz] += QQQ0[qz][qy][qx] * Gt[l][k] * s;
+                  v[qz] += QQQ1[qz][qy][qx] * Bt[j][i];
+                  w[qz] += QQQ2[qz][qy][qx] * Bt[j][i];
+               }
+            }
+            MFEM_UNROLL(MQ1)
+            for (int qz = 0; qz < Q1D; ++qz)
+            {
+               QQD0[qz][qy][dx] = u[qz];
+               QQD1[qz][qy][dx] = v[qz];
+               QQD2[qz][qy][dx] = w[qz];
+            }
+         }
+      }
+      MFEM_SYNC_THREAD;
+      MFEM_FOREACH_THREAD(dy,y,D1D)
+      {
+         MFEM_FOREACH_THREAD(dx,x,D1D)
+         {
+            double u[Q1D], v[Q1D], w[Q1D];
+            MFEM_UNROLL(MQ1)
+            for (int qz = 0; qz < Q1D; ++qz) { u[qz] = v[qz] = w[qz] = 0.0; }
+            MFEM_UNROLL(MQ1)
+            for (int qy = 0; qy < Q1D; ++qy)
+            {
+               const int i = qi(qy,dy,Q1D);
+               const int j = dj(qy,dy,D1D);
+               const int k = qk(qy,dy,Q1D);
+               const int l = dl(qy,dy,D1D);
+               const double s = sign(qy,dy);
+               MFEM_UNROLL(MQ1)
+               for (int qz = 0; qz < Q1D; ++qz)
+               {
+                  u[qz] += QQD0[qz][qy][dx] * Bt[j][i];
+                  v[qz] += QQD1[qz][qy][dx] * Gt[l][k] * s;
+                  w[qz] += QQD2[qz][qy][dx] * Bt[j][i];
+               }
+            }
+            MFEM_UNROLL(MQ1)
+            for (int qz = 0; qz < Q1D; ++qz)
+            {
+               QDD0[qz][dy][dx] = u[qz];
+               QDD1[qz][dy][dx] = v[qz];
+               QDD2[qz][dy][dx] = w[qz];
+            }
+         }
+      }
+      MFEM_SYNC_THREAD;
+      MFEM_FOREACH_THREAD(dy,y,D1D)
+      {
+         MFEM_FOREACH_THREAD(dx,x,D1D)
+         {
+            double u[D1D], v[D1D], w[D1D];
+            MFEM_UNROLL(MD1)
+            for (int dz = 0; dz < D1D; ++dz) { u[dz] = v[dz] = w[dz] = 0.0; }
+            MFEM_UNROLL(MQ1)
+            for (int qz = 0; qz < Q1D; ++qz)
+            {
+               MFEM_UNROLL(MD1)
+               for (int dz = 0; dz < D1D; ++dz)
               {
                  const int i = qi(qz,dz,Q1D);
                  const int j = dj(qz,dz,D1D);
                  const int k = qk(qz,dz,Q1D);
                  const int l = dl(qz,dz,D1D);
                  const double s = sign(qz,dz);
-                  u += QDD0[qz][dy][dx] * Bt[j][i];
-                  v += QDD1[qz][dy][dx] * Bt[j][i];
-                  w += QDD2[qz][dy][dx] * Gt[l][k] * s;
+                  u[dz] += QDD0[qz][dy][dx] * Bt[j][i];
+                  v[dz] += QDD1[qz][dy][dx] * Bt[j][i];
+                  w[dz] += QDD2[qz][dy][dx] * Gt[l][k] * s;
               }
-               y(dx,dy,dz,e) += (u + v + w);
+            }
+            MFEM_UNROLL(MD1)
+            for (int dz = 0; dz < D1D; ++dz)
+            {
+               y(dx,dy,dz,e) += (u[dz] + v[dz] + w[dz]);
            }
         }
      }
@@ -1845,7 +1877,6 @@ static void PADiffusionApply(const int dim,
   {
      switch (ID)
      {
-         case 0x22: return SmemPADiffusionApply3D<2,2>(NE,symm,B,G,D,X,Y);
         case 0x23: return SmemPADiffusionApply3D<2,3>(NE,symm,B,G,D,X,Y);
         case 0x34: return SmemPADiffusionApply3D<3,4>(NE,symm,B,G,D,X,Y);
         case 0x45: return SmemPADiffusionApply3D<4,5>(NE,symm,B,G,D,X,Y);
@@ -21,12 +21,12 @@ namespace mfem
 // PA Divergence Integrator

 // PA Divergence Assemble 2D kernel
-static void PADivergenceSetup2D(const int Q1D,
-                                const int NE,
-                                const Array<double> &w,
-                                const Vector &j,
-                                const double COEFF,
-                                Vector &op)
+void PADivergenceSetup2D(const int Q1D,
+                         const int NE,
+                         const Array<double> &w,
+                         const Vector &j,
+                         const double COEFF,
+                         Vector &op)
 {
   const int NQ = Q1D*Q1D;
   auto W = w.Read();
@@ -51,12 +51,12 @@ static void PADivergenceSetup2D(const int Q1D,
 }

 // PA Divergence Assemble 3D kernel
-static void PADivergenceSetup3D(const int Q1D,
-                                const int NE,
-                                const Array<double> &w,
-                                const Vector &j,
-                                const double COEFF,
-                                Vector &op)
+void PADivergenceSetup3D(const int Q1D,
+                         const int NE,
+                         const Array<double> &w,
+                         const Vector &j,
+                         const double COEFF,
+                         Vector &op)
 {
   const int NQ = Q1D*Q1D*Q1D;
   auto W = w.Read();
@@ -160,16 +160,16 @@ void VectorDivergenceIntegrator::AssemblePA(const FiniteElementSpace &trial_fes,

 // PA Divergence Apply 2D kernel
 template<const int T_TR_D1D = 0, const int T_TE_D1D = 0, const int T_Q1D = 0>
-static void PADivergenceApply2D(const int NE,
-                                const Array<double> &b,
-                                const Array<double> &g,
-                                const Array<double> &bt,
-                                const Vector &op_,
-                                const Vector &x_,
-                                Vector &y_,
-                                const int tr_d1d = 0,
-                                const int te_d1d = 0,
-                                const int q1d = 0)
+void PADivergenceApply2D(const int NE,
+                         const Array<double> &b,
+                         const Array<double> &g,
+                         const Array<double> &bt,
+                         const Vector &op_,
+                         const Vector &x_,
+                         Vector &y_,
+                         const int tr_d1d = 0,
+                         const int te_d1d = 0,
+                         const int q1d = 0)
 {
   const int TR_D1D = T_TR_D1D ? T_TR_D1D : tr_d1d;
   const int TE_D1D = T_TE_D1D ? T_TE_D1D : te_d1d;
@@ -281,16 +281,16 @@ static void PADivergenceApply2D(const int NE,
 // Shared memory PA Divergence Apply 2D kernel
 template<const int T_TR_D1D = 0, const int T_TE_D1D = 0, const int T_Q1D = 0,
         const int T_NBZ = 0>
-static void SmemPADivergenceApply2D(const int NE,
-                                    const Array<double> &b_,
-                                    const Array<double> &g_,
-                                    const Array<double> &bt_,
-                                    const Vector &op_,
-                                    const Vector &x_,
-                                    Vector &y_,
-                                    const int tr_d1d = 0,
-                                    const int te_d1d = 0,
-                                    const int q1d = 0)
+void SmemPADivergenceApply2D(const int NE,
+                             const Array<double> &b_,
+                             const Array<double> &g_,
+                             const Array<double> &bt_,
+                             const Vector &op_,
+                             const Vector &x_,
+                             Vector &y_,
+                             const int tr_d1d = 0,
+                             const int te_d1d = 0,
+                             const int q1d = 0)
 {
   // TODO
   MFEM_ASSERT(false, "SHARED MEM NOT PROGRAMMED YET");
@@ -298,16 +298,16 @@ static void SmemPADivergenceApply2D(const int NE,

 // PA Divergence Apply 2D kernel transpose
 template<const int T_TR_D1D = 0, const int T_TE_D1D = 0, const int T_Q1D = 0>
-static void PADivergenceApplyTranspose2D(const int NE,
-                                         const Array<double> &bt,
-                                         const Array<double> &gt,
-                                         const Array<double> &b,
-                                         const Vector &op_,
-                                         const Vector &x_,
-                                         Vector &y_,
-                                         const int tr_d1d = 0,
-                                         const int te_d1d = 0,
-                                         const int q1d = 0)
+void PADivergenceApplyTranspose2D(const int NE,
+                                  const Array<double> &bt,
+                                  const Array<double> &gt,
+                                  const Array<double> &b,
+                                  const Vector &op_,
+                                  const Vector &x_,
+                                  Vector &y_,
+                                  const int tr_d1d = 0,
+                                  const int te_d1d = 0,
+                                  const int q1d = 0)
 {
   const int TR_D1D = T_TR_D1D ? T_TR_D1D : tr_d1d;
   const int TE_D1D = T_TE_D1D ? T_TE_D1D : te_d1d;
@@ -414,16 +414,16 @@ static void PADivergenceApplyTranspose2D(const int NE,

 // PA Vector Divergence Apply 3D kernel
 template<const int T_TR_D1D = 0, const int T_TE_D1D = 0, const int T_Q1D = 0>
-static void PADivergenceApply3D(const int NE,
-                                const Array<double> &b,
-                                const Array<double> &g,
-                                const Array<double> &bt,
-                                const Vector &op_,
-                                const Vector &x_,
-                                Vector &y_,
-                                int tr_d1d = 0,
-                                int te_d1d = 0,
-                                int q1d = 0)
+void PADivergenceApply3D(const int NE,
+                         const Array<double> &b,
+                         const Array<double> &g,
+                         const Array<double> &bt,
+                         const Vector &op_,
+                         const Vector &x_,
+                         Vector &y_,
+                         int tr_d1d = 0,
+                         int te_d1d = 0,
+                         int q1d = 0)
 {
   const int TR_D1D = T_TR_D1D ? T_TR_D1D : tr_d1d;
   const int TE_D1D = T_TE_D1D ? T_TE_D1D : te_d1d;
@@ -597,16 +597,16 @@ static void PADivergenceApply3D(const int NE,

 // PA Vector Divergence Apply 3D kernel
 template<const int T_TR_D1D = 0, const int T_TE_D1D = 0, const int T_Q1D = 0>
-static void PADivergenceApplyTranspose3D(const int NE,
-                                         const Array<double> &bt,
-                                         const Array<double> &gt,
-                                         const Array<double> &b,
-                                         const Vector &op_,
-                                         const Vector &x_,
-                                         Vector &y_,
-                                         int tr_d1d = 0,
-                                         int te_d1d = 0,
-                                         int q1d = 0)
+void PADivergenceApplyTranspose3D(const int NE,
+                                  const Array<double> &bt,
+                                  const Array<double> &gt,
+                                  const Array<double> &b,
+                                  const Vector &op_,
+                                  const Vector &x_,
+                                  Vector &y_,
+                                  int tr_d1d = 0,
+                                  int te_d1d = 0,
+                                  int q1d = 0)
 {
   const int TR_D1D = T_TR_D1D ? T_TR_D1D : tr_d1d;
   const int TE_D1D = T_TE_D1D ? T_TE_D1D : te_d1d;
@@ -775,16 +775,16 @@ static void PADivergenceApplyTranspose3D(const int NE,

 // Shared memory PA Vector Divergence Apply 3D kernel
 template<const int T_TR_D1D = 0, const int T_TE_D1D = 0, const int T_Q1D = 0>
-static void SmemPADivergenceApply3D(const int NE,
-                                    const Array<double> &b_,
-                                    const Array<double> &g_,
-                                    const Array<double> &bt_,
-                                    const Vector &q_,
-                                    const Vector &x_,
-                                    Vector &y_,
-                                    const int tr_d1d = 0,
-                                    const int te_d1d = 0,
-                                    const int q1d = 0)
+void SmemPADivergenceApply3D(const int NE,
+                             const Array<double> &b_,
+                             const Array<double> &g_,
+                             const Array<double> &bt_,
+                             const Vector &q_,
+                             const Vector &x_,
+                             Vector &y_,
+                             const int tr_d1d = 0,
+                             const int te_d1d = 0,
+                             const int q1d = 0)
 {
   const int TR_D1D = T_TR_D1D ? T_TR_D1D : tr_d1d;
   const int TE_D1D = T_TE_D1D ? T_TE_D1D : te_d1d;
@@ -70,12 +70,12 @@ namespace mfem
   the \b MFEM_SHARED keyword for local arrays. */

 // PA Gradient Assemble 2D kernel
-static void PAGradientSetup2D(const int Q1D,
-                              const int NE,
-                              const Array<double> &w,
-                              const Vector &j,
-                              const Vector &c,
-                              Vector &op)
+void PAGradientSetup2D(const int Q1D,
+                       const int NE,
+                       const Array<double> &w,
+                       const Vector &j,
+                       const Vector &c,
+                       Vector &op)
 {
   const int NQ = Q1D*Q1D;
   auto W = w.Read();
@@ -105,12 +105,12 @@ static void PAGradientSetup2D(const int Q1D,
 }

 // PA Gradient Assemble 3D kernel
-static void PAGradientSetup3D(const int Q1D,
-                              const int NE,
-                              const Array<double> &w,
-                              const Vector &j,
-                              const Vector &c,
-                              Vector &op)
+void PAGradientSetup3D(const int Q1D,
+                       const int NE,
+                       const Array<double> &w,
+                       const Vector &j,
+                       const Vector &c,
+                       Vector &op)
 {
   const int NQ = Q1D*Q1D*Q1D;
   auto W = w.Read();
@@ -254,16 +254,16 @@ void GradientIntegrator::AssemblePA(const FiniteElementSpace &trial_fes,

 // PA Gradient Apply 2D kernel
 template<int T_TR_D1D = 0, int T_TE_D1D = 0, int T_Q1D = 0>
-static void PAGradientApply2D(const int NE,
-                              const Array<double> &b,
-                              const Array<double> &g,
-                              const Array<double> &bt,
-                              const Vector &op_,
-                              const Vector &x_,
-                              Vector &y_,
-                              const int tr_d1d = 0,
-                              const int te_d1d = 0,
-                              const int q1d = 0)
+void PAGradientApply2D(const int NE,
+                       const Array<double> &b,
+                       const Array<double> &g,
+                       const Array<double> &bt,
+                       const Vector &op_,
+                       const Vector &x_,
+                       Vector &y_,
+                       const int tr_d1d = 0,
+                       const int te_d1d = 0,
+                       const int q1d = 0)
 {
   const int TR_D1D = T_TR_D1D ? T_TR_D1D : tr_d1d;
   const int TE_D1D = T_TE_D1D ? T_TE_D1D : te_d1d;
@@ -384,16 +384,16 @@ static void PAGradientApplyTranspose2D(const int NE,

 // PA Gradient Apply 3D kernel
 template<const int T_TR_D1D = 0, const int T_TE_D1D = 0, const int T_Q1D = 0>
-static void PAGradientApply3D(const int NE,
-                              const Array<double> &b,
-                              const Array<double> &g,
-                              const Array<double> &bt,
-                              const Vector &op_,
-                              const Vector &x_,
-                              Vector &y_,
-                              int tr_d1d = 0,
-                              int te_d1d = 0,
-                              int q1d = 0)
+void PAGradientApply3D(const int NE,
+                       const Array<double> &b,
+                       const Array<double> &g,
+                       const Array<double> &bt,
+                       const Vector &op_,
+                       const Vector &x_,
+                       Vector &y_,
+                       int tr_d1d = 0,
+                       int te_d1d = 0,
+                       int q1d = 0)
 {
   const int TR_D1D = T_TR_D1D ? T_TR_D1D : tr_d1d;
   const int TE_D1D = T_TE_D1D ? T_TE_D1D : te_d1d;
@@ -579,16 +579,16 @@ static void PAGradientApplyTranspose3D(const int NE,

 // Shared memory PA Gradient Apply 3D kernel
 template<const int T_TR_D1D = 0, const int T_TE_D1D = 0, const int T_Q1D = 0>
-static void SmemPAGradientApply3D(const int NE,
-                                  const Array<double> &b_,
-                                  const Array<double> &g_,
-                                  const Array<double> &bt_,
-                                  const Vector &d_,
-                                  const Vector &x_,
-                                  Vector &y_,
-                                  const int tr_d1d = 0,
-                                  const int te_d1d = 0,
-                                  const int q1d = 0)
+void SmemPAGradientApply3D(const int NE,
+                           const Array<double> &b_,
+                           const Array<double> &g_,
+                           const Array<double> &bt_,
+                           const Vector &d_,
+                           const Vector &x_,
+                           Vector &y_,
+                           const int tr_d1d = 0,
+                           const int te_d1d = 0,
+                           const int q1d = 0)
 {
   const int TR_D1D = T_TR_D1D ? T_TR_D1D : tr_d1d;
   const int TE_D1D = T_TE_D1D ? T_TE_D1D : te_d1d;
@@ -186,7 +186,7 @@ void PAHcurlMassAssembleDiagonal2D(const int D1D,
                  const double wy = (c == 1) ? Bo(qy,dy) : Bc(qy,dy);

                  mass[qx] += wy * wy * ((c == 0) ? op(qx,qy,0,e) :
-                                         op(qx,qy,symmetric ? 2 : 3, e));
+                  op(qx,qy,symmetric ? 2 : 3, e));
               }
            }

@@ -237,7 +237,7 @@ void PAHcurlMassAssembleDiagonal3D(const int D1D,
         const int D1Dx = (c == 0) ? D1D - 1 : D1D;

         const int opc = (c == 0) ? 0 : ((c == 1) ? (symmetric ? 3 : 4) :
-                                         (symmetric ? 5 : 8));
+         (symmetric ? 5 : 8));

         double mass[MAX_Q1D];

@@ -791,12 +791,12 @@ void SmemPAHcurlMassApply3D(const int D1D,
 }

 // PA H(curl) curl-curl assemble 2D kernel
-static void PACurlCurlSetup2D(const int Q1D,
-                              const int NE,
-                              const Array<double> &w,
-                              const Vector &j,
-                              Vector &coeff,
-                              Vector &op)
+void PACurlCurlSetup2D(const int Q1D,
+                       const int NE,
+                       const Array<double> &w,
+                       const Vector &j,
+                       Vector &coeff,
+                       Vector &op)
 {
   const int NQ = Q1D*Q1D;
   auto W = w.Read();
@@ -818,13 +818,13 @@ static void PACurlCurlSetup2D(const int Q1D,
 }

 // PA H(curl) curl-curl assemble 3D kernel
-static void PACurlCurlSetup3D(const int Q1D,
-                              const int coeffDim,
-                              const int NE,
-                              const Array<double> &w,
-                              const Vector &j,
-                              Vector &coeff,
-                              Vector &op)
+void PACurlCurlSetup3D(const int Q1D,
+                       const int coeffDim,
+                       const int NE,
+                       const Array<double> &w,
+                       const Vector &j,
+                       Vector &coeff,
+                       Vector &op)
 {
   const int NQ = Q1D*Q1D*Q1D;
   const bool symmetric = (coeffDim != 9);
@@ -1045,16 +1045,16 @@ void CurlCurlIntegrator::AssemblePA(const FiniteElementSpace &fes)
   }
 }

-static void PACurlCurlApply2D(const int D1D,
-                              const int Q1D,
-                              const int NE,
-                              const Array<double> &bo,
-                              const Array<double> &bot,
-                              const Array<double> &gc,
-                              const Array<double> &gct,
-                              const Vector &pa_data,
-                              const Vector &x,
-                              Vector &y)
+void PACurlCurlApply2D(const int D1D,
+                       const int Q1D,
+                       const int NE,
+                       const Array<double> &bo,
+                       const Array<double> &bot,
+                       const Array<double> &gc,
+                       const Array<double> &gct,
+                       const Vector &pa_data,
+                       const Vector &x,
+                       Vector &y)
 {
   constexpr static int VDIM = 2;
   constexpr static int MAX_D1D = HCURL_MAX_D1D;
@@ -1166,19 +1166,19 @@ static void PACurlCurlApply2D(const int D1D,
 }

 template<int MAX_D1D = HCURL_MAX_D1D, int MAX_Q1D = HCURL_MAX_Q1D>
-static void PACurlCurlApply3D(const int D1D,
-                              const int Q1D,
-                              const bool symmetric,
-                              const int NE,
-                              const Array<double> &bo,
-                              const Array<double> &bc,
-                              const Array<double> &bot,
-                              const Array<double> &bct,
-                              const Array<double> &gc,
-                              const Array<double> &gct,
-                              const Vector &pa_data,
-                              const Vector &x,
-                              Vector &y)
+void PACurlCurlApply3D(const int D1D,
+                       const int Q1D,
+                       const bool symmetric,
+                       const int NE,
+                       const Array<double> &bo,
+                       const Array<double> &bc,
+                       const Array<double> &bot,
+                       const Array<double> &bct,
+                       const Array<double> &gc,
+                       const Array<double> &gct,
+                       const Vector &pa_data,
+                       const Vector &x,
+                       Vector &y)
 {
   MFEM_VERIFY(D1D <= MAX_D1D, "Error: D1D > MAX_D1D");
   MFEM_VERIFY(Q1D <= MAX_Q1D, "Error: Q1D > MAX_Q1D");
@@ -1677,19 +1677,19 @@ static void PACurlCurlApply3D(const int D1D,
 }

 template<int MAX_D1D = HCURL_MAX_D1D, int MAX_Q1D = HCURL_MAX_Q1D>
-static void SmemPACurlCurlApply3D(const int D1D,
-                                  const int Q1D,
-                                  const bool symmetric,
-                                  const int NE,
-                                  const Array<double> &bo,
-                                  const Array<double> &bc,
-                                  const Array<double> &bot,
-                                  const Array<double> &bct,
-                                  const Array<double> &gc,
-                                  const Array<double> &gct,
-                                  const Vector &pa_data,
-                                  const Vector &x,
-                                  Vector &y)
+void SmemPACurlCurlApply3D(const int D1D,
+                           const int Q1D,
+                           const bool symmetric,
+                           const int NE,
+                           const Array<double> &bo,
+                           const Array<double> &bc,
+                           const Array<double> &bot,
+                           const Array<double> &bct,
+                           const Array<double> &gc,
+                           const Array<double> &gct,
+                           const Vector &pa_data,
+                           const Vector &x,
+                           Vector &y)
 {
   MFEM_VERIFY(D1D <= MAX_D1D, "Error: D1D > MAX_D1D");
   MFEM_VERIFY(Q1D <= MAX_Q1D, "Error: Q1D > MAX_Q1D");
@@ -2032,13 +2032,13 @@ void CurlCurlIntegrator::AddMultPA(const Vector &x, Vector &y) const
   }
 }

-static void PACurlCurlAssembleDiagonal2D(const int D1D,
-                                         const int Q1D,
-                                         const int NE,
-                                         const Array<double> &bo,
-                                         const Array<double> &gc,
-                                         const Vector &pa_data,
-                                         Vector &diag)
+void PACurlCurlAssembleDiagonal2D(const int D1D,
+                                  const int Q1D,
+                                  const int NE,
+                                  const Array<double> &bo,
+                                  const Array<double> &gc,
+                                  const Vector &pa_data,
+                                  Vector &diag)
 {
   constexpr static int VDIM = 2;
   constexpr static int MAX_Q1D = HCURL_MAX_Q1D;
@@ -2087,16 +2087,16 @@ static void PACurlCurlAssembleDiagonal2D(const int D1D,
 }

 template<int MAX_D1D = HCURL_MAX_D1D, int MAX_Q1D = HCURL_MAX_Q1D>
-static void PACurlCurlAssembleDiagonal3D(const int D1D,
-                                         const int Q1D,
-                                         const bool symmetric,
-                                         const int NE,
-                                         const Array<double> &bo,
-                                         const Array<double> &bc,
-                                         const Array<double> &go,
-                                         const Array<double> &gc,
-                                         const Vector &pa_data,
-                                         Vector &diag)
+void PACurlCurlAssembleDiagonal3D(const int D1D,
+                                  const int Q1D,
+                                  const bool symmetric,
+                                  const int NE,
+                                  const Array<double> &bo,
+                                  const Array<double> &bc,
+                                  const Array<double> &go,
+                                  const Array<double> &gc,
+                                  const Vector &pa_data,
+                                  Vector &diag)
 {
   constexpr static int VDIM = 3;
   MFEM_VERIFY(D1D <= MAX_D1D, "Error: D1D > MAX_D1D");
@@ -2273,16 +2273,16 @@ static void PACurlCurlAssembleDiagonal3D(const int D1D,
 }

 template<int MAX_D1D = HCURL_MAX_D1D, int MAX_Q1D = HCURL_MAX_Q1D>
-static void SmemPACurlCurlAssembleDiagonal3D(const int D1D,
-                                             const int Q1D,
-                                             const bool symmetric,
-                                             const int NE,
-                                             const Array<double> &bo,
-                                             const Array<double> &bc,
-                                             const Array<double> &go,
-                                             const Array<double> &gc,
-                                             const Vector &pa_data,
-                                             Vector &diag)
+void SmemPACurlCurlAssembleDiagonal3D(const int D1D,
+                                      const int Q1D,
+                                      const bool symmetric,
+                                      const int NE,
+                                      const Array<double> &bo,
+                                      const Array<double> &bc,
+                                      const Array<double> &go,
+                                      const Array<double> &gc,
+                                      const Vector &pa_data,
+                                      Vector &diag)
 {
   MFEM_VERIFY(D1D <= MAX_D1D, "Error: D1D > MAX_D1D");
   MFEM_VERIFY(Q1D <= MAX_Q1D, "Error: Q1D > MAX_Q1D");
@@ -2955,18 +2955,18 @@ void MixedVectorCurlIntegrator::AssemblePA(const FiniteElementSpace &trial_fes,
 // Apply to x corresponding to DOF's in H(curl) (trial), whose curl is
 // integrated against H(curl) test functions corresponding to y.
 template<int MAX_D1D = HCURL_MAX_D1D, int MAX_Q1D = HCURL_MAX_Q1D>
-static void PAHcurlL2Apply3D(const int D1D,
-                             const int Q1D,
-                             const int coeffDim,
-                             const int NE,
-                             const Array<double> &bo,
-                             const Array<double> &bc,
-                             const Array<double> &bot,
-                             const Array<double> &bct,
-                             const Array<double> &gc,
-                             const Vector &pa_data,
-                             const Vector &x,
-                             Vector &y)
+void PAHcurlL2Apply3D(const int D1D,
+                      const int Q1D,
+                      const int coeffDim,
+                      const int NE,
+                      const Array<double> &bo,
+                      const Array<double> &bc,
+                      const Array<double> &bot,
+                      const Array<double> &bct,
+                      const Array<double> &gc,
+                      const Vector &pa_data,
+                      const Vector &x,
+                      Vector &y)
 {
   MFEM_VERIFY(D1D <= MAX_D1D, "Error: D1D > MAX_D1D");
   MFEM_VERIFY(Q1D <= MAX_Q1D, "Error: Q1D > MAX_Q1D");
@@ -3297,16 +3297,16 @@ static void PAHcurlL2Apply3D(const int D1D,
 // Apply to x corresponding to DOF's in H(curl) (trial), whose curl is
 // integrated against H(curl) test functions corresponding to y.
 template<int MAX_D1D = HCURL_MAX_D1D, int MAX_Q1D = HCURL_MAX_Q1D>
-static void SmemPAHcurlL2Apply3D(const int D1D,
-                                 const int Q1D,
-                                 const int coeffDim,
-                                 const int NE,
-                                 const Array<double> &bo,
-                                 const Array<double> &bc,
-                                 const Array<double> &gc,
-                                 const Vector &pa_data,
-                                 const Vector &x,
-                                 Vector &y)
+void SmemPAHcurlL2Apply3D(const int D1D,
+                          const int Q1D,
+                          const int coeffDim,
+                          const int NE,
+                          const Array<double> &bo,
+                          const Array<double> &bc,
+                          const Array<double> &gc,
+                          const Vector &pa_data,
+                          const Vector &x,
+                          Vector &y)
 {
   MFEM_VERIFY(D1D <= MAX_D1D, "Error: D1D > MAX_D1D");
   MFEM_VERIFY(Q1D <= MAX_Q1D, "Error: Q1D > MAX_Q1D");
@@ -3585,18 +3585,18 @@ static void SmemPAHcurlL2Apply3D(const int D1D,
 // Apply to x corresponding to DOF's in H(curl) (trial), whose curl is
 // integrated against H(div) test functions corresponding to y.
 template<int MAX_D1D = HCURL_MAX_D1D, int MAX_Q1D = HCURL_MAX_Q1D>
-static void PAHcurlHdivApply3D(const int D1D,
-                               const int D1Dtest,
-                               const int Q1D,
-                               const int NE,
-                               const Array<double> &bo,
-                               const Array<double> &bc,
-                               const Array<double> &bot,
-                               const Array<double> &bct,
-                               const Array<double> &gc,
-                               const Vector &pa_data,
-                               const Vector &x,
-                               Vector &y)
+void PAHcurlHdivApply3D(const int D1D,
+                        const int D1Dtest,
+                        const int Q1D,
+                        const int NE,
+                        const Array<double> &bo,
+                        const Array<double> &bc,
+                        const Array<double> &bot,
+                        const Array<double> &bct,
+                        const Array<double> &gc,
+                        const Vector &pa_data,
+                        const Vector &x,
+                        Vector &y)
 {
   MFEM_VERIFY(D1D <= MAX_D1D, "Error: D1D > MAX_D1D");
   MFEM_VERIFY(Q1D <= MAX_Q1D, "Error: Q1D > MAX_Q1D");
@@ -4071,18 +4071,18 @@ void MixedVectorWeakCurlIntegrator::AssemblePA(const FiniteElementSpace
 // Apply to x corresponding to DOF's in H(curl) (trial), integrated against curl
 // of H(curl) test functions corresponding to y.
 template<int MAX_D1D = HCURL_MAX_D1D, int MAX_Q1D = HCURL_MAX_Q1D>
-static void PAHcurlL2Apply3DTranspose(const int D1D,
-                                      const int Q1D,
-                                      const int coeffDim,
-                                      const int NE,
-                                      const Array<double> &bo,
-                                      const Array<double> &bc,
-                                      const Array<double> &bot,
-                                      const Array<double> &bct,
-                                      const Array<double> &gct,
-                                      const Vector &pa_data,
-                                      const Vector &x,
-                                      Vector &y)
+void PAHcurlL2Apply3DTranspose(const int D1D,
+                               const int Q1D,
+                               const int coeffDim,
+                               const int NE,
+                               const Array<double> &bo,
+                               const Array<double> &bc,
+                               const Array<double> &bot,
+                               const Array<double> &bct,
+                               const Array<double> &gct,
+                               const Vector &pa_data,
+                               const Vector &x,
+                               Vector &y)
 {
   // See PAHcurlL2Apply3D for comments.

@@ -4413,16 +4413,16 @@ static void PAHcurlL2Apply3DTranspose(const int D1D,
 }

 template<int MAX_D1D = HCURL_MAX_D1D, int MAX_Q1D = HCURL_MAX_Q1D>
-static void SmemPAHcurlL2Apply3DTranspose(const int D1D,
-                                          const int Q1D,
-                                          const int coeffDim,
-                                          const int NE,
-                                          const Array<double> &bo,
-                                          const Array<double> &bc,
-                                          const Array<double> &gc,
-                                          const Vector &pa_data,
-                                          const Vector &x,
-                                          Vector &y)
+void SmemPAHcurlL2Apply3DTranspose(const int D1D,
+                                   const int Q1D,
+                                   const int coeffDim,
+                                   const int NE,
+                                   const Array<double> &bo,
+                                   const Array<double> &bc,
+                                   const Array<double> &gc,
+                                   const Vector &pa_data,
+                                   const Vector &x,
+                                   Vector &y)
 {
   MFEM_VERIFY(D1D <= MAX_D1D, "Error: D1D > MAX_D1D");
   MFEM_VERIFY(Q1D <= MAX_Q1D, "Error: Q1D > MAX_Q1D");
@@ -4675,13 +4675,13 @@ void MixedVectorWeakCurlIntegrator::AddMultPA(const Vector &x, Vector &y) const
 // Apply to x corresponding to DOFs in H^1 (domain) the (topological) gradient
 // to get a dof in H(curl) (range). You can think of the range as the "test" space
 // and the domain as the "trial" space, but there's no integration.
-static void PAHcurlApplyGradient2D(const int c_dofs1D,
-                                   const int o_dofs1D,
-                                   const int NE,
-                                   const Array<double> &B_,
-                                   const Array<double> &G_,
-                                   const Vector &x_,
-                                   Vector &y_)
+void PAHcurlApplyGradient2D(const int c_dofs1D,
+                            const int o_dofs1D,
+                            const int NE,
+                            const Array<double> &B_,
+                            const Array<double> &G_,
+                            const Vector &x_,
+                            Vector &y_)
 {
   auto B = Reshape(B_.Read(), c_dofs1D, c_dofs1D);
   auto G = Reshape(G_.Read(), o_dofs1D, c_dofs1D);
@@ -4753,12 +4753,12 @@ static void PAHcurlApplyGradient2D(const int c_dofs1D,
 }

 // Specialization of PAHcurlApplyGradient2D to the case where B is identity
-static void PAHcurlApplyGradient2DBId(const int c_dofs1D,
-                                      const int o_dofs1D,
-                                      const int NE,
-                                      const Array<double> &G_,
-                                      const Vector &x_,
-                                      Vector &y_)
+void PAHcurlApplyGradient2DBId(const int c_dofs1D,
+                               const int o_dofs1D,
+                               const int NE,
+                               const Array<double> &G_,
+                               const Vector &x_,
+                               Vector &y_)
 {
   auto G = Reshape(G_.Read(), o_dofs1D, c_dofs1D);

@@ -4822,7 +4822,7 @@ static void PAHcurlApplyGradient2DBId(const int c_dofs1D,
   });
 }

-static void PAHcurlApplyGradientTranspose2D(
+void PAHcurlApplyGradientTranspose2D(
   const int c_dofs1D, const int o_dofs1D, const int NE,
   const Array<double> &B_, const Array<double> &G_,
   const Vector &x_, Vector &y_)
@@ -4898,7 +4898,7 @@ static void PAHcurlApplyGradientTranspose2D(

 // Specialization of PAHcurlApplyGradientTranspose2D to the case where
 // B is identity
-static void PAHcurlApplyGradientTranspose2DBId(
+void PAHcurlApplyGradientTranspose2DBId(
   const int c_dofs1D, const int o_dofs1D, const int NE,
   const Array<double> &G_,
   const Vector &x_, Vector &y_)
@@ -4965,13 +4965,13 @@ static void PAHcurlApplyGradientTranspose2DBId(
   });
 }

-static void PAHcurlApplyGradient3D(const int c_dofs1D,
-                                   const int o_dofs1D,
-                                   const int NE,
-                                   const Array<double> &B_,
-                                   const Array<double> &G_,
-                                   const Vector &x_,
-                                   Vector &y_)
+void PAHcurlApplyGradient3D(const int c_dofs1D,
+                            const int o_dofs1D,
+                            const int NE,
+                            const Array<double> &B_,
+                            const Array<double> &G_,
+                            const Vector &x_,
+                            Vector &y_)
 {
   auto B = Reshape(B_.Read(), c_dofs1D, c_dofs1D);
   auto G = Reshape(G_.Read(), o_dofs1D, c_dofs1D);
@@ -5154,12 +5154,12 @@ static void PAHcurlApplyGradient3D(const int c_dofs1D,
 }

 // Specialization of PAHcurlApplyGradient3D to the case where
-static void PAHcurlApplyGradient3DBId(const int c_dofs1D,
-                                      const int o_dofs1D,
-                                      const int NE,
-                                      const Array<double> &G_,
-                                      const Vector &x_,
-                                      Vector &y_)
+void PAHcurlApplyGradient3DBId(const int c_dofs1D,
+                               const int o_dofs1D,
+                               const int NE,
+                               const Array<double> &G_,
+                               const Vector &x_,
+                               Vector &y_)
 {
   auto G = Reshape(G_.Read(), o_dofs1D, c_dofs1D);

@@ -5322,7 +5322,7 @@ static void PAHcurlApplyGradient3DBId(const int c_dofs1D,
   });
 }

-static void PAHcurlApplyGradientTranspose3D(
+void PAHcurlApplyGradientTranspose3D(
   const int c_dofs1D, const int o_dofs1D, const int NE,
   const Array<double> &B_, const Array<double> &G_,
   const Vector &x_, Vector &y_)
@@ -5507,7 +5507,7 @@ static void PAHcurlApplyGradientTranspose3D(
 }

 // Specialization of PAHcurlApplyGradientTranspose3D to the case where
-static void PAHcurlApplyGradientTranspose3DBId(
+void PAHcurlApplyGradientTranspose3DBId(
   const int c_dofs1D, const int o_dofs1D, const int NE,
   const Array<double> &G_,
   const Vector &x_, Vector &y_)
@@ -5789,14 +5789,14 @@ void GradientInterpolator::AddMultTransposePA(const Vector &x, Vector &y) const
   }
 }

-static void PAHcurlVecH1IdentityApply3D(const int c_dofs1D,
-                                        const int o_dofs1D,
-                                        const int NE,
-                                        const Array<double> &Bclosed,
-                                        const Array<double> &Bopen,
-                                        const Vector &pa_data,
-                                        const Vector &x_,
-                                        Vector &y_)
+void PAHcurlVecH1IdentityApply3D(const int c_dofs1D,
+                                 const int o_dofs1D,
+                                 const int NE,
+                                 const Array<double> &Bclosed,
+                                 const Array<double> &Bopen,
+                                 const Vector &pa_data,
+                                 const Vector &x_,
+                                 Vector &y_)
 {
   auto Bc = Reshape(Bclosed.Read(), c_dofs1D, c_dofs1D);
   auto Bo = Reshape(Bopen.Read(), o_dofs1D, c_dofs1D);
@@ -6002,14 +6002,14 @@ static void PAHcurlVecH1IdentityApply3D(const int c_dofs1D,
   });
 }

-static void PAHcurlVecH1IdentityApplyTranspose3D(const int c_dofs1D,
-                                                 const int o_dofs1D,
-                                                 const int NE,
-                                                 const Array<double> &Bclosed,
-                                                 const Array<double> &Bopen,
-                                                 const Vector &pa_data,
-                                                 const Vector &x_,
-                                                 Vector &y_)
+void PAHcurlVecH1IdentityApplyTranspose3D(const int c_dofs1D,
+                                          const int o_dofs1D,
+                                          const int NE,
+                                          const Array<double> &Bclosed,
+                                          const Array<double> &Bopen,
+                                          const Vector &pa_data,
+                                          const Vector &x_,
+                                          Vector &y_)
 {
   auto Bc = Reshape(Bclosed.Read(), c_dofs1D, c_dofs1D);
   auto Bo = Reshape(Bopen.Read(), o_dofs1D, c_dofs1D);
@@ -6228,14 +6228,14 @@ static void PAHcurlVecH1IdentityApplyTranspose3D(const int c_dofs1D,
   });
 }

-static void PAHcurlVecH1IdentityApply2D(const int c_dofs1D,
-                                        const int o_dofs1D,
-                                        const int NE,
-                                        const Array<double> &Bclosed,
-                                        const Array<double> &Bopen,
-                                        const Vector &pa_data,
-                                        const Vector &x_,
-                                        Vector &y_)
+void PAHcurlVecH1IdentityApply2D(const int c_dofs1D,
+                                 const int o_dofs1D,
+                                 const int NE,
+                                 const Array<double> &Bclosed,
+                                 const Array<double> &Bopen,
+                                 const Vector &pa_data,
+                                 const Vector &x_,
+                                 Vector &y_)
 {
   auto Bc = Reshape(Bclosed.Read(), c_dofs1D, c_dofs1D);
   auto Bo = Reshape(Bopen.Read(), o_dofs1D, c_dofs1D);
@@ -6327,14 +6327,14 @@ static void PAHcurlVecH1IdentityApply2D(const int c_dofs1D,
   });
 }

-static void PAHcurlVecH1IdentityApplyTranspose2D(const int c_dofs1D,
-                                                 const int o_dofs1D,
-                                                 const int NE,
-                                                 const Array<double> &Bclosed,
-                                                 const Array<double> &Bopen,
-                                                 const Vector &pa_data,
-                                                 const Vector &x_,
-                                                 Vector &y_)
+void PAHcurlVecH1IdentityApplyTranspose2D(const int c_dofs1D,
+                                          const int o_dofs1D,
+                                          const int NE,
+                                          const Array<double> &Bclosed,
+                                          const Array<double> &Bopen,
+                                          const Vector &pa_data,
+                                          const Vector &x_,
+                                          Vector &y_)
 {
   auto Bc = Reshape(Bclosed.Read(), c_dofs1D, c_dofs1D);
   auto Bo = Reshape(Bopen.Read(), o_dofs1D, c_dofs1D);
@@ -539,12 +539,12 @@ void PAHdivMassApply3D(const int D1D,

 // PA H(div) div-div assemble 2D kernel
 // NOTE: this is identical to PACurlCurlSetup3D
-static void PADivDivSetup2D(const int Q1D,
-                            const int NE,
-                            const Array<double> &w,
-                            const Vector &j,
-                            Vector &coeff_,
-                            Vector &op)
+void PADivDivSetup2D(const int Q1D,
+                     const int NE,
+                     const Array<double> &w,
+                     const Vector &j,
+                     Vector &coeff_,
+                     Vector &op)
 {
   const int NQ = Q1D*Q1D;
   auto W = w.Read();
@@ -565,12 +565,12 @@ static void PADivDivSetup2D(const int Q1D,
   });
 }

-static void PADivDivSetup3D(const int Q1D,
-                            const int NE,
-                            const Array<double> &w,
-                            const Vector &j,
-                            Vector &coeff_,
-                            Vector &op)
+void PADivDivSetup3D(const int Q1D,
+                     const int NE,
+                     const Array<double> &w,
+                     const Vector &j,
+                     Vector &coeff_,
+                     Vector &op)
 {
   const int NQ = Q1D*Q1D*Q1D;
   auto W = w.Read();
@@ -599,16 +599,16 @@ static void PADivDivSetup3D(const int Q1D,
   });
 }

-static void PADivDivApply2D(const int D1D,
-                            const int Q1D,
-                            const int NE,
-                            const Array<double> &Bo_,
-                            const Array<double> &Gc_,
-                            const Array<double> &Bot_,
-                            const Array<double> &Gct_,
-                            const Vector &op_,
-                            const Vector &x_,
-                            Vector &y_)
+void PADivDivApply2D(const int D1D,
+                     const int Q1D,
+                     const int NE,
+                     const Array<double> &Bo_,
+                     const Array<double> &Gc_,
+                     const Array<double> &Bot_,
+                     const Array<double> &Gct_,
+                     const Vector &op_,
+                     const Vector &x_,
+                     Vector &y_)
 {
   constexpr static int VDIM = 2;
   constexpr static int MAX_D1D = HDIV_MAX_D1D;
@@ -718,16 +718,16 @@ static void PADivDivApply2D(const int D1D,
   }); // end of element loop
 }

-static void PADivDivApply3D(const int D1D,
-                            const int Q1D,
-                            const int NE,
-                            const Array<double> &Bo_,
-                            const Array<double> &Gc_,
-                            const Array<double> &Bot_,
-                            const Array<double> &Gct_,
-                            const Vector &op_,
-                            const Vector &x_,
-                            Vector &y_)
+void PADivDivApply3D(const int D1D,
+                     const int Q1D,
+                     const int NE,
+                     const Array<double> &Bo_,
+                     const Array<double> &Gc_,
+                     const Array<double> &Bot_,
+                     const Array<double> &Gct_,
+                     const Vector &op_,
+                     const Vector &x_,
+                     Vector &y_)
 {
   MFEM_VERIFY(D1D <= HDIV_MAX_D1D, "Error: D1D > HDIV_MAX_D1D");
   MFEM_VERIFY(Q1D <= HDIV_MAX_Q1D, "Error: Q1D > HDIV_MAX_Q1D");
@@ -967,13 +967,13 @@ void DivDivIntegrator::AddMultPA(const Vector &x, Vector &y) const
   }
 }

-static void PADivDivAssembleDiagonal2D(const int D1D,
-                                       const int Q1D,
-                                       const int NE,
-                                       const Array<double> &Bo_,
-                                       const Array<double> &Gc_,
-                                       const Vector &op_,
-                                       Vector &diag_)
+void PADivDivAssembleDiagonal2D(const int D1D,
+                                const int Q1D,
+                                const int NE,
+                                const Array<double> &Bo_,
+                                const Array<double> &Gc_,
+                                const Vector &op_,
+                                Vector &diag_)
 {
   constexpr static int VDIM = 2;
   constexpr static int MAX_Q1D = HDIV_MAX_Q1D;
@@ -1023,13 +1023,13 @@ static void PADivDivAssembleDiagonal2D(const int D1D,
   });
 }

-static void PADivDivAssembleDiagonal3D(const int D1D,
-                                       const int Q1D,
-                                       const int NE,
-                                       const Array<double> &Bo_,
-                                       const Array<double> &Gc_,
-                                       const Vector &op_,
-                                       Vector &diag_)
+void PADivDivAssembleDiagonal3D(const int D1D,
+                                const int Q1D,
+                                const int NE,
+                                const Array<double> &Bo_,
+                                const Array<double> &Gc_,
+                                const Vector &op_,
+                                Vector &diag_)
 {
   MFEM_VERIFY(D1D <= HDIV_MAX_D1D, "Error: D1D > HDIV_MAX_D1D");
   MFEM_VERIFY(Q1D <= HDIV_MAX_Q1D, "Error: Q1D > HDIV_MAX_Q1D");
@@ -1104,11 +1104,11 @@ void DivDivIntegrator::AssembleDiagonalPA(Vector& diag)
 }

 // PA H(div)-L2 (div u, p) assemble 2D kernel
-static void PADivL2Setup2D(const int Q1D,
-                           const int NE,
-                           const Array<double> &w,
-                           Vector &coeff_,
-                           Vector &op)
+void PADivL2Setup2D(const int Q1D,
+                    const int NE,
+                    const Array<double> &w,
+                    Vector &coeff_,
+                    Vector &op)
 {
   const int NQ = Q1D*Q1D;
   auto W = w.Read();
@@ -1123,11 +1123,11 @@ static void PADivL2Setup2D(const int Q1D,
   });
 }

-static void PADivL2Setup3D(const int Q1D,
-                           const int NE,
-                           const Array<double> &w,
-                           Vector &coeff_,
-                           Vector &op)
+void PADivL2Setup3D(const int Q1D,
+                    const int NE,
+                    const Array<double> &w,
+                    Vector &coeff_,
+                    Vector &op)
 {
   const int NQ = Q1D*Q1D*Q1D;
   auto W = w.Read();
@@ -1225,16 +1225,16 @@ VectorFEDivergenceIntegrator::AssemblePA(const FiniteElementSpace &trial_fes,

 // Apply to x corresponding to DOF's in H(div) (trial), whose divergence is
 // integrated against L_2 test functions corresponding to y.
-static void PAHdivL2Apply3D(const int D1D,
-                            const int Q1D,
-                            const int L2D1D,
-                            const int NE,
-                            const Array<double> &Bo_,
-                            const Array<double> &Gc_,
-                            const Array<double> &L2Bot_,
-                            const Vector &op_,
-                            const Vector &x_,
-                            Vector &y_)
+void PAHdivL2Apply3D(const int D1D,
+                     const int Q1D,
+                     const int L2D1D,
+                     const int NE,
+                     const Array<double> &Bo_,
+                     const Array<double> &Gc_,
+                     const Array<double> &L2Bot_,
+                     const Vector &op_,
+                     const Vector &x_,
+                     Vector &y_)
 {
   MFEM_VERIFY(D1D <= HDIV_MAX_D1D, "Error: D1D > HDIV_MAX_D1D");
   MFEM_VERIFY(Q1D <= HDIV_MAX_Q1D, "Error: Q1D > HDIV_MAX_Q1D");
@@ -1388,16 +1388,16 @@ static void PAHdivL2Apply3D(const int D1D,

 // Apply to x corresponding to DOF's in H(div) (trial), whose divergence is
 // integrated against L_2 test functions corresponding to y.
-static void PAHdivL2Apply2D(const int D1D,
-                            const int Q1D,
-                            const int L2D1D,
-                            const int NE,
-                            const Array<double> &Bo_,
-                            const Array<double> &Gc_,
-                            const Array<double> &L2Bot_,
-                            const Vector &op_,
-                            const Vector &x_,
-                            Vector &y_)
+void PAHdivL2Apply2D(const int D1D,
+                     const int Q1D,
+                     const int L2D1D,
+                     const int NE,
+                     const Array<double> &Bo_,
+                     const Array<double> &Gc_,
+                     const Array<double> &L2Bot_,
+                     const Vector &op_,
+                     const Vector &x_,
+                     Vector &y_)
 {
   constexpr static int VDIM = 2;
   constexpr static int MAX_D1D = HDIV_MAX_D1D;
@@ -1494,16 +1494,16 @@ static void PAHdivL2Apply2D(const int D1D,
   }); // end of element loop
 }

-static void PAHdivL2ApplyTranspose3D(const int D1D,
-                                     const int Q1D,
-                                     const int L2D1D,
-                                     const int NE,
-                                     const Array<double> &L2Bo_,
-                                     const Array<double> &Gct_,
-                                     const Array<double> &Bot_,
-                                     const Vector &op_,
-                                     const Vector &x_,
-                                     Vector &y_)
+void PAHdivL2ApplyTranspose3D(const int D1D,
+                              const int Q1D,
+                              const int L2D1D,
+                              const int NE,
+                              const Array<double> &L2Bo_,
+                              const Array<double> &Gct_,
+                              const Array<double> &Bot_,
+                              const Vector &op_,
+                              const Vector &x_,
+                              Vector &y_)
 {
   MFEM_VERIFY(D1D <= HDIV_MAX_D1D, "Error: D1D > HDIV_MAX_D1D");
   MFEM_VERIFY(Q1D <= HDIV_MAX_Q1D, "Error: Q1D > HDIV_MAX_Q1D");
@@ -1656,16 +1656,16 @@ static void PAHdivL2ApplyTranspose3D(const int D1D,
   }); // end of element loop
 }

-static void PAHdivL2ApplyTranspose2D(const int D1D,
-                                     const int Q1D,
-                                     const int L2D1D,
-                                     const int NE,
-                                     const Array<double> &L2Bo_,
-                                     const Array<double> &Gct_,
-                                     const Array<double> &Bot_,
-                                     const Vector &op_,
-                                     const Vector &x_,
-                                     Vector &y_)
+void PAHdivL2ApplyTranspose2D(const int D1D,
+                              const int Q1D,
+                              const int L2D1D,
+                              const int NE,
+                              const Array<double> &L2Bo_,
+                              const Array<double> &Gct_,
+                              const Array<double> &Bot_,
+                              const Vector &op_,
+                              const Vector &x_,
+                              Vector &y_)
 {
   constexpr static int VDIM = 2;
   constexpr static int MAX_D1D = HDIV_MAX_D1D;
@@ -1791,16 +1791,16 @@ void VectorFEDivergenceIntegrator::AddMultTransposePA(const Vector &x,
   }
 }

-static void PAHdivL2AssembleDiagonal_ADAt_3D(const int D1D,
-                                             const int Q1D,
-                                             const int L2D1D,
-                                             const int NE,
-                                             const Array<double> &L2Bo_,
-                                             const Array<double> &Gct_,
-                                             const Array<double> &Bot_,
-                                             const Vector &op_,
-                                             const Vector &D_,
-                                             Vector &diag_)
+void PAHdivL2AssembleDiagonal_ADAt_3D(const int D1D,
+                                      const int Q1D,
+                                      const int L2D1D,
+                                      const int NE,
+                                      const Array<double> &L2Bo_,
+                                      const Array<double> &Gct_,
+                                      const Array<double> &Bot_,
+                                      const Vector &op_,
+                                      const Vector &D_,
+                                      Vector &diag_)
 {
   MFEM_VERIFY(D1D <= HDIV_MAX_D1D, "Error: D1D > HDIV_MAX_D1D");
   MFEM_VERIFY(Q1D <= HDIV_MAX_Q1D, "Error: Q1D > HDIV_MAX_Q1D");
@@ -1916,16 +1916,16 @@ static void PAHdivL2AssembleDiagonal_ADAt_3D(const int D1D,
   }); // end of element loop
 }

-static void PAHdivL2AssembleDiagonal_ADAt_2D(const int D1D,
-                                             const int Q1D,
-                                             const int L2D1D,
-                                             const int NE,
-                                             const Array<double> &L2Bo_,
-                                             const Array<double> &Gct_,
-                                             const Array<double> &Bot_,
-                                             const Vector &op_,
-                                             const Vector &D_,
-                                             Vector &diag_)
+void PAHdivL2AssembleDiagonal_ADAt_2D(const int D1D,
+                                      const int Q1D,
+                                      const int L2D1D,
+                                      const int NE,
+                                      const Array<double> &L2Bo_,
+                                      const Array<double> &Gct_,
+                                      const Array<double> &Bot_,
+                                      const Vector &op_,
+                                      const Vector &D_,
+                                      Vector &diag_)
 {
   constexpr static int VDIM = 2;

@@ -17,13 +17,13 @@ namespace mfem
 {

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EAMassAssemble1D(const int NE,
-                             const Array<double> &basis,
-                             const Vector &padata,
-                             Vector &eadata,
-                             const bool add,
-                             const int d1d = 0,
-                             const int q1d = 0)
+void EAMassAssemble1D(const int NE,
+                      const Array<double> &basis,
+                      const Vector &padata,
+                      Vector &eadata,
+                      const bool add,
+                      const int d1d = 0,
+                      const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -67,13 +67,13 @@ static void EAMassAssemble1D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EAMassAssemble2D(const int NE,
-                             const Array<double> &basis,
-                             const Vector &padata,
-                             Vector &eadata,
-                             const bool add,
-                             const int d1d = 0,
-                             const int q1d = 0)
+void EAMassAssemble2D(const int NE,
+                      const Array<double> &basis,
+                      const Vector &padata,
+                      Vector &eadata,
+                      const bool add,
+                      const int d1d = 0,
+                      const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -139,13 +139,13 @@ static void EAMassAssemble2D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void EAMassAssemble3D(const int NE,
-                             const Array<double> &basis,
-                             const Vector &padata,
-                             Vector &eadata,
-                             const bool add,
-                             const int d1d = 0,
-                             const int q1d = 0)
+void EAMassAssemble3D(const int NE,
+                      const Array<double> &basis,
+                      const Vector &padata,
+                      Vector &eadata,
+                      const bool add,
+                      const int d1d = 0,
+                      const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -155,12 +155,12 @@ void MassIntegrator::AssemblePA(const FiniteElementSpace &fes)
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void PAMassAssembleDiagonal2D(const int NE,
-                                     const Array<double> &b,
-                                     const Vector &d,
-                                     Vector &y,
-                                     const int d1d = 0,
-                                     const int q1d = 0)
+void PAMassAssembleDiagonal2D(const int NE,
+                              const Array<double> &b,
+                              const Vector &d,
+                              Vector &y,
+                              const int d1d = 0,
+                              const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -201,12 +201,12 @@ static void PAMassAssembleDiagonal2D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0, int T_NBZ = 0>
-static void SmemPAMassAssembleDiagonal2D(const int NE,
-                                         const Array<double> &b_,
-                                         const Vector &d_,
-                                         Vector &y_,
-                                         const int d1d = 0,
-                                         const int q1d = 0)
+void SmemPAMassAssembleDiagonal2D(const int NE,
+                                  const Array<double> &b_,
+                                  const Vector &d_,
+                                  Vector &y_,
+                                  const int d1d = 0,
+                                  const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -267,12 +267,12 @@ static void SmemPAMassAssembleDiagonal2D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void PAMassAssembleDiagonal3D(const int NE,
-                                     const Array<double> &b,
-                                     const Vector &d,
-                                     Vector &y,
-                                     const int d1d = 0,
-                                     const int q1d = 0)
+void PAMassAssembleDiagonal3D(const int NE,
+                              const Array<double> &b,
+                              const Vector &d,
+                              Vector &y,
+                              const int d1d = 0,
+                              const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -336,12 +336,12 @@ static void PAMassAssembleDiagonal3D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void SmemPAMassAssembleDiagonal3D(const int NE,
-                                         const Array<double> &b_,
-                                         const Vector &d_,
-                                         Vector &y_,
-                                         const int d1d = 0,
-                                         const int q1d = 0)
+void SmemPAMassAssembleDiagonal3D(const int NE,
+                                  const Array<double> &b_,
+                                  const Vector &d_,
+                                  Vector &y_,
+                                  const int d1d = 0,
+                                  const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -569,14 +569,14 @@ static void OccaPAMassApply3D(const int D1D,
 #endif // MFEM_USE_OCCA

 template<int T_D1D = 0, int T_Q1D = 0>
-static void PAMassApply2D(const int NE,
-                          const Array<double> &b_,
-                          const Array<double> &bt_,
-                          const Vector &d_,
-                          const Vector &x_,
-                          Vector &y_,
-                          const int d1d = 0,
-                          const int q1d = 0)
+void PAMassApply2D(const int NE,
+                   const Array<double> &b_,
+                   const Array<double> &bt_,
+                   const Vector &d_,
+                   const Vector &x_,
+                   Vector &y_,
+                   const int d1d = 0,
+                   const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -661,14 +661,14 @@ static void PAMassApply2D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0, int T_NBZ = 0>
-static void SmemPAMassApply2D(const int NE,
-                              const Array<double> &b_,
-                              const Array<double> &bt_,
-                              const Vector &d_,
-                              const Vector &x_,
-                              Vector &y_,
-                              const int d1d = 0,
-                              const int q1d = 0)
+void SmemPAMassApply2D(const int NE,
+                       const Array<double> &b_,
+                       const Array<double> &bt_,
+                       const Vector &d_,
+                       const Vector &x_,
+                       Vector &y_,
+                       const int d1d = 0,
+                       const int q1d = 0)
 {
   MFEM_CONTRACT_VAR(bt_);
   const int D1D = T_D1D ? T_D1D : d1d;
@@ -784,14 +784,14 @@ static void SmemPAMassApply2D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void PAMassApply3D(const int NE,
-                          const Array<double> &b_,
-                          const Array<double> &bt_,
-                          const Vector &d_,
-                          const Vector &x_,
-                          Vector &y_,
-                          const int d1d = 0,
-                          const int q1d = 0)
+void PAMassApply3D(const int NE,
+                   const Array<double> &b_,
+                   const Array<double> &bt_,
+                   const Vector &d_,
+                   const Vector &x_,
+                   Vector &y_,
+                   const int d1d = 0,
+                   const int q1d = 0)
 {
   const int D1D = T_D1D ? T_D1D : d1d;
   const int Q1D = T_Q1D ? T_Q1D : q1d;
@@ -925,14 +925,14 @@ static void PAMassApply3D(const int NE,
 }

 template<int T_D1D = 0, int T_Q1D = 0>
-static void SmemPAMassApply3D(const int NE,
-                              const Array<double> &b_,
-                              const Array<double> &bt_,
-                              const Vector &d_,
-                              const Vector &x_,
-                              Vector &y_,
-                              const int d1d = 0,
-                              const int q1d = 0)
+void SmemPAMassApply3D(const int NE,
+                       const Array<double> &b_,
+                       const Array<double> &bt_,
+                       const Vector &d_,
+                       const Vector &x_,
+                       Vector &y_,
+                       const int d1d = 0,
+                       const int q1d = 0)
 {
   MFEM_CONTRACT_VAR(bt_);
   const int D1D = T_D1D ? T_D1D : d1d;
@@ -1203,10 +1203,8 @@ static void PAMassApply(const int dim,
   {
      switch (id)
      {
-         case 0x22: return SmemPAMassApply3D<2,2>(NE,B,Bt,D,X,Y);
         case 0x23: return SmemPAMassApply3D<2,3>(NE,B,Bt,D,X,Y);
         case 0x24: return SmemPAMassApply3D<2,4>(NE,B,Bt,D,X,Y);
-         case 0x26: return SmemPAMassApply3D<2,6>(NE,B,Bt,D,X,Y);
         case 0x34: return SmemPAMassApply3D<3,4>(NE,B,Bt,D,X,Y);
         case 0x35: return SmemPAMassApply3D<3,5>(NE,B,Bt,D,X,Y);
         case 0x36: return SmemPAMassApply3D<3,6>(NE,B,Bt,D,X,Y);
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Tzanio	967908029f	make style	2021-09-13 10:51:52 -07:00
$AMCBRIDGE\ahudym$ AMCBRIDGE\ahudym	8560248ea4	revert make file	2021-09-13 18:42:10 +03:00
$AMCBRIDGE\ahudym$ AMCBRIDGE\ahudym	fa85cf2471	fix style	2021-09-13 18:41:43 +03:00
$AMCBRIDGE\ahudym$ AMCBRIDGE\ahudym	8b34881d46	change #elif with #else	2021-09-13 11:39:22 +03:00
$AMCBRIDGE\ahudym$ AMCBRIDGE\ahudym	de0cad550d	cast to SOCKET only for WIN32 platform	2021-09-13 11:34:44 +03:00
$AMCBRIDGE\ahudym$ AMCBRIDGE\ahudym	1f7f84d7a2	1. Remove "static" keyword from functions used as lambdas on "device" 2. Add _USE_MATH_DEFINES to CmakeLists.txt (without this change VS doesn't "see" this preprocessor in config.hpp) 3. Remove asm definition - asm is not allowed in VS anymore 4. Convert "const double Epsilon" to constexpr (fixes CUDA compilation error)	2021-09-03 11:59:51 +03:00