Commit Graph

Select branches

Hide Pull Requests

2.0

3.0

3.1

3.2

3.3

3.4

5.0

fixSignedZerosWithFastMath

master

Mono Color

• 12c9c14bc5 CMake/CI: Cache the CUDA test builds with ccache master 3 people 2026-08-21 19:27:36 -07:00
• 16e143a970 GPU: Add a unified-memory storage benchmark for integrated GPUs (NVIDIA Jetson) Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-21 13:23:56 -07:00
• eb4ea17496 Core: Add a PPC Clang optimization barrier Florian Maurin 2026-08-21 19:57:15 +00:00
• a474202651 SME: Add a double-precision GEMM kernel, and fix four streaming-mode pessimizations Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-21 09:43:52 -07:00
• 28d5232c8f GPU: Do not assert in noexcept move assignment Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-21 09:42:01 -07:00
• c9030f29e7 Core: Fix complex exp edge cases on modern MSVC Florian Maurin 2026-08-21 16:16:26 +00:00
• 8763a05898 Core: Avoid NaN seeds in RVV min/max reductions Florian Maurin 2026-08-21 16:13:09 +00:00
• b1f4d24c8a GPU: Bind dense solver host input through Ref instead of copying (found on NVIDIA Jetson AGX Orin) Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-20 23:05:06 -07:00
• 083d44a902 BDCSVD: Split the bidiagonal at negligible superdiagonal entries Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-20 22:39:45 -07:00
• 85d8e2effb CI: Run the emulated test jobs on the largest hosted runner Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-20 22:34:57 -07:00
• 199dfbb6fa GPU: Honor the host outer stride when uploading matrices (found on NVIDIA Jetson AGX Orin) Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-20 20:50:57 -07:00
• 61f550a350 Core: Fix ambiguous half construction under nvcc on arm64 (found on NVIDIA Jetson AGX Orin) Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-20 20:18:39 -07:00
• 4456c1f559 GPU: Skip the remaining GPU tests when no CUDA device is present Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-20 19:02:30 -07:00
• a36ff9c948 Core: Preserve generic Clang masks under fast-math Florian Maurin 2026-08-21 00:36:36 +00:00
• dd03dd029b GPU: Skip the device_matrix test when no CUDA device is present Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-20 17:03:44 -07:00
• 2a76726d2a GPU: Warm up the GPU benchmarks before timing them (found on NVIDIA Jetson AGX Orin) Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-20 16:57:14 -07:00
• d812864302 BLAS: Pass string length in xerbla_ calls for Fortran ABI compatibility Antonio Sánchez 2026-08-20 23:47:03 +00:00
• 9be565330f CI: Add an affected-tests tier between smoke and all-tests 3 people 2026-08-20 15:45:06 -07:00
• ce99aad6c5 Tensor: Avoid integer overflow in reverse block tests Florian Maurin 2026-08-20 22:41:16 +00:00
• 4b3097dcc1 StructuredMatrices: Tie conjugate-pair moduli in the Circulant SVD Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-20 11:48:13 -07:00
• 54a71ce153 GPU: Measure benchmark rate counters against wall time Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-20 00:18:20 -07:00
• d376536929 Core: Use NumTraits::epsilon() in the nextafter test Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-19 21:51:01 -07:00
• d5d9cb636f CI: Skip the GPU module in the clang-tidy job Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-19 21:29:17 -07:00
• 57bb412d55 StructuredMatrices: Stop using the deprecated SVD constructor Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-19 21:12:26 -07:00
• bd919651d0 Core: Fix AVX-512 trsm no-malloc fallback using the wrong panel origin Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-19 20:51:56 -07:00
• 74ce7e6d59 Householder: Preserve reflectors for large components 3 people 2026-08-19 16:24:16 -07:00
• c095d7e78e Core: Make PropagateNaN min/max agree on signed-zero ties Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-19 15:18:17 -07:00
• 09401d580f CI: Check each changed source once with clang-tidy 3 people 2026-08-19 13:30:23 -07:00
• e82accca02 Tensor: Fix narrowing-index tests on ILP32 targets Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-19 08:29:31 -07:00
• 303c4c54db Core: Use a scalar tail for inner-unrolled partial-packet assignments Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-19 08:17:54 -07:00
• 7c8275ef88 FFT: Serialize FFTW plan creation and destruction Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-18 22:34:03 -07:00
• 74e7795b6a Tensor: Serve striding packets from inner runs Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-18 22:29:16 -07:00
• d9799b408a Sparse: Fix AmbiVector element lifetimes and a per-product Scalar leak Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-18 22:02:30 -07:00
• a38efc01c8 Householder: Avoid shifting by bool Florian Maurin 2026-08-19 04:24:28 +00:00
• e80109341d Householder: Preserve significant small vector tails Florian Maurin 2026-08-19 03:32:11 +00:00
• f752349f69 CMake: Declare the project version in project() Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-18 19:28:48 -07:00
• 4e57e79dd8 Splines: Let KnotAveraging accept parameters outside the unit interval Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-18 19:24:51 -07:00
• 79a228e372 CMake: Add EIGEN_INSTALL to disable Eigen's install rules Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-18 19:12:45 -07:00
• 18f650509b CMake: Ground the documented CMake behavior in tests Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-18 19:02:52 -07:00
• cd5b44a5ad Tensor: Serve write-side blocks for concat, reverse and roll destinations Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-18 18:55:46 -07:00
• 2759846aaa Tensor: Fix index type handling in variadic accessors Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-18 18:49:41 -07:00
• a885abcfa7 Tensor: Copy unit-magnitude reversed runs with a packet reversal Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-18 16:53:39 -07:00
• 976f7e3d2a Core: Support SSE partial loads on older GCC Florian Maurin 2026-08-18 23:51:37 +00:00
• 17ccff85e0 Tensor: Fix GCC ICE in complex reductions Florian Maurin 2026-08-18 23:28:28 +00:00
• 8d5ffc2d53 AltiVec: Let the GEBP kernels take a narrower index type Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-18 12:54:40 -07:00
• 0192017d2c FFT: Accept matrices as one-dimensional operands Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 21:48:05 -07:00
• 2d22fabad0 LevenbergMarquardt: Fix the sparse Levenberg-Marquardt subproblem solve Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 21:24:25 -07:00
• 657e192bfc Geometry: Add Hyperplane::isCoincident Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 21:16:38 -07:00
• 43d03664d1 Optimize NEON 64-bit integral packets onalante-ebay 2026-08-18 04:11:32 +00:00
• 5593b98453 Core: Implement numext::nextafter for all floating point types Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 21:01:28 -07:00
• e857597ca0 Docs: Correct three claims the manual audit left behind Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 20:52:31 -07:00
• e70059f86a SpecialFunctions: Restore erf/erfc for custom scalars and long double Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 19:32:40 -07:00
• b0d40c93fc QR: Speed up the BQRRP sketch update in RandColPivHouseholderQR Charles Schlosser 2026-08-18 02:22:48 +00:00
• be74d8961e Core: Do not promote an unwound Block to an inner panel Charles Schlosser 2026-08-18 02:07:44 +00:00
• 4f19d2ba4a QR: Accept a row-major MatrixType in RandColPivHouseholderQR Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 17:25:25 -07:00
• 11a02f2908 Docs: Audit and refresh the user manual under doc/ Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 17:16:09 -07:00
• 5e585f7e3a SME: Stream SYMM packers and vectorize mapper fallbacks Kasper Nielsen 2026-08-17 23:37:51 +00:00
• 04e6354681 Core: Vectorize the SYMM packers by delegating to the gemm packers Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 16:30:19 -07:00
• 76a7c53018 Core: Make structured-bindings tuple_size SFINAE-friendly for dynamic sizes Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 10:12:24 -07:00
• 14d33f2f2e Tests: Derive strided-Map outer strides from outerStride() Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 08:42:35 -07:00
• b31f97c897 NonLinearOptimization: Fix r1mpyq() for row-major storage Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 08:41:19 -07:00
• b4093b1bbb Eigenvalues: Stop GeneralizedSelfAdjointEigenSolver::compute from allocating Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 01:42:38 -07:00
• 4aa66c4ed6 Core: Do not inherit RowMajorBit into a vector-shaped IndexedView Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-17 00:33:10 -07:00
• c4236f4a60 Doc: Correct the EIGEN_STACK_ALLOCATION_LIMIT=0 semantics Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-16 19:54:22 -07:00
• f5494736fd Doc: State that TriangularBase coefficient access is raw storage access Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-16 18:59:01 -07:00
• e453d30949 Doc: Add the missing decompositions to the dense catalogue Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-16 18:53:52 -07:00
• bbac95e9ec Tensor: Derive customOp output rank from the functor's dimensions() Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-16 17:44:41 -07:00
• 8cef20fd23 Core: Fix SSE2 pcmp_lt<Packet2l> comparing low halves as signed Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-16 17:39:42 -07:00
• 8b40a356b9 Tensor: Serve blocks from the patch-family evaluators Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-16 17:34:55 -07:00
• 20590ecaab StructuredMatrices: Fix Doxygen \overload usage breaking the docs build Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-16 17:27:14 -07:00
• 919847505f StructuredMatrices: Add KroneckerOperator implicit fast operator 3 people 2026-08-16 16:01:50 -07:00
• 11ae29bc32 Cholesky: Fix LDLT::rankUpdate dropping updates and misreporting status Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-16 14:51:40 -07:00
• 24e7b41e1b Tensor: Make SwappedLayout capturable-free for MSVC in block-eval test Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-15 23:15:57 -07:00
• 19e57e021e CI: Skip tests whose binaries already passed in earlier MR pipelines Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-15 23:01:53 -07:00
• 004a031d2a CI: Stop re-compressing the ccache payload when archiving Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-15 22:58:34 -07:00
• f59d68ea77 CI: Size the ccache caches to hold a full build Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-15 22:13:03 -07:00
• 8fdb9a609a CI: Find Visual Studio Build Tools installs with vswhere -products * Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-15 20:33:46 -07:00
• 9acad8e42b Core: Modernize and speed up the generic clang vector backend Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-15 18:31:54 -07:00
• a0edc3964b Core: Name the Array shift operators for the shift they perform Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-15 18:19:50 -07:00
• 9b1c027e6d Core: Fix integer packet shifts on AltiVec and AVX2 Florian Maurin 2026-08-16 01:03:39 +00:00
• be15280b77 Core: Use __builtin_elementwise_{minnum,maxnum} for float packets Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-15 12:52:42 -07:00
• 4ac0b07cbc Tensor: Serve blocks from the strided-slice evaluators Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-15 12:44:01 -07:00
• cb7353bfd8 Tensor: Serve layout-swap blocks beyond raw buffers Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-15 12:29:40 -07:00
• 45da26b764 Core: Share GEMV wrapper utilities and fix a triangular RHS buffer leak Florian Maurin 2026-08-15 17:37:28 +00:00
• 47324cc2af Fix four defects found by auditing Doxygen claims against the code Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-14 19:38:35 -07:00
• ddd2165ab4 Fix four defects found by auditing Doxygen claims against the code Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-14 15:13:12 -07:00
• f0501ba630 Cleanup: Move SFINAE constraints to template parameters Florian Maurin 2026-08-14 21:08:27 +00:00
• 24c57ef1fc Core: Replace recursive utility expansions Florian Maurin 2026-08-14 18:32:12 +00:00
• 07612387bb Core: Simplify callable detection Florian Maurin 2026-08-14 18:22:25 +00:00
• 5ee913f028 Fix unsafe use of EIGEN_ALIGN_MAX for aligned-packet data onalante-ebay 2026-08-14 17:01:08 +00:00
• 5210c47d7d Test: Skip the subnormal stableNormalize check where FTZ flushes it Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-14 08:02:22 -07:00
• 05def587c8 Tensor: Normalize cv-qualified index queries Florian Maurin 2026-08-14 14:43:26 +00:00
• 410fa16a77 Docs: Correct documented claims that the code contradicts Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-14 07:33:02 -07:00
• 5bdc386dab Use numext:: math in library headers Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-14 07:32:26 -07:00
• 59208527c5 Core: Restore the configuration macros the typedef sweep expanded Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-13 20:42:46 -07:00
• 973c210150 Test: Accept ulp-level differences in the bicgstab issue-2856 regression Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-13 19:51:45 -07:00
• 996e56744a CI: Extend the ccache build cache to the Windows runners Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-13 18:29:37 -07:00
• 6d035ce6b1 Docs: Add sparse agent guidance and document the blocking docs build Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-13 17:30:41 -07:00
• 4529102e6d Replace typedef with using in Eigen/src Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-13 15:33:04 -07:00
• 3b5d23ecb2 Replace NULL with nullptr Rasmus Munk LarsenandRasmus Munk Larsen 2026-08-13 15:32:03 -07:00