phynexis-deps/OpenBLAS
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
10,034 Commits 60 Branches 70 Tags
eb9bbcc299c79423b6d9461ba3ee1f2a47ccaa2d
Commit Graph
342 Commits
Author SHA1 Message Date
Martin Kroeker 55b16e5923 Merge pull request #5643 from martin-frbg/neov2_param 
Fix SGEMM returning wrong results in multithreading on NeoverseV2
 2026-03-06 11:51:08 +01:00
Fadi Arafeh f30202b705 Accelerate SVE128 SBGEMM/BGEMM 
This accelerates SBGEMM/BGEMM by extending the existing 8x4 kernel to 8x8 (unrolling N by 8)

Not sure if it's a good idea to delete the previous 8x4 kernel?

Here are the speedups on single core Neoverse-V2 (SVE128) compared to prev state:

Per-shape speedup
  M=N=K=64: SBGEMM 1.164x (16.42%), BGEMM 1.133x (13.30%)
  M=N=K=128: SBGEMM 1.220x (22.02%), BGEMM 1.186x (18.56%)
  M=N=K=256: SBGEMM 1.241x (24.08%), BGEMM 1.235x (23.54%)
  M=N=K=512: SBGEMM 1.240x (23.95%), BGEMM 1.227x (22.75%)
  M=N=K=1024: SBGEMM 1.251x (25.11%), BGEMM 1.232x (23.23%)
  M=N=K=2048: SBGEMM 1.235x (23.47%), BGEMM 1.246x (24.64%)

Signed-off-by: Fadi Arafeh <fadi.arafeh@arm.com>
 2026-03-05 13:50:07 +00:00
Martin Kroeker db00d5c2c9 Fix SGEMM returning wrong results in multithreading on NeoverseV2 2026-02-12 00:02:13 +01:00
Martin Kroeker 7ffce1c788 fix spurious change of (S)BGEMM parameters for NeoverseV1 2026-01-14 19:03:06 +01:00
Martin Kroeker 6de062cfc2 Merge branch 'OpenMathLib:develop' into issue5414 2026-01-11 17:45:11 +01:00
Martin Kroeker 52ec7faf31 Merge pull request #5554 from hideaki-motoki/issue5553_gemm_default_pqr_for_a64fx 
Setting optimized `[SD]GEMM_DEFAULT_[PQR]` parameters for `A64FX`
 2026-01-11 15:59:51 +01:00
Martin Kroeker 2d46f1ec65 Merge branch 'develop' into issue5414 2026-01-09 15:04:06 +01:00
Martin Kroeker c040d5ed86 Merge pull request #5591 from quic/topic/ssyr2k_direct_sme1 
Support for SME1 based ssyr2k_direct kernel for cblas_ssyr2k level 3 API
 2026-01-08 15:47:38 +01:00
Zhiqing xie 6939a43c3b Support for SME1 based ssyr2k_direct kernel for cblas_ssyr2k level 3 API 2026-01-08 11:09:04 +08:00
Martin Kroeker 83a788c387 Add BLASLONG cast to the DEFAULT_ALIGN parameter of Cooper Lake and Sapphire Rapids 2025-12-29 23:23:34 +01:00
h-motoki 5f0735832b fix param.h: turn [sd]gemm_default_[pqr] parameters for a64fx 2025-11-28 13:27:23 +09:00
Martin Kroeker ea85b6696f Merge branch 'OpenMathLib:develop' into issue5414 2025-11-23 10:14:07 +01:00
Martin Kroeker f2d010de12 Merge pull request #5512 from quic/topic/ssyrk_direct_sme1 
Support for SME1 based ssyrk_direct kernel for cblas_ssyrk level 3 API
 2025-11-06 14:06:43 -08:00
changjua 43d38d336f Support for SME1 based ssyrk_direct kernel for cblas_ssyrk level 3 API 2025-10-20 11:35:20 +08:00
Martin Kroeker d7b0fccbb4 Enable SME-based kernels for VortexM4 with clang-based compilers only 2025-10-19 13:34:26 -07:00
Martin Kroeker 9bfc3612f9 Merge branch 'OpenMathLib:develop' into issue5414 2025-10-12 09:18:06 -07:00
Martin Kroeker e40714cabd Merge pull request #5450 from quic/topic/strmm_direct_sme1 
Support for SME1 based strmm_direct kernel for cblas_strmm level 3 API
 2025-10-11 15:20:19 -07:00
changjua 644ea07ef9 Support for SME1 based strmm_direct kernel for cblas_strmm level 3 API 2025-10-10 10:48:27 +08:00
Martin Kroeker fc516af155 Merge branch 'develop' into issue5414 2025-10-01 14:12:59 -07:00
Martin Kroeker e939c6c315 Merge pull request #5471 from quic/topic/ssymm_direct_sme1 
Support for SME1 based ssymm_direct kernel for cblas_ssymm level 3 API
 2025-10-01 06:22:36 -07:00
Rajendra Prasad Matcha 19268471cc Support for SME1 based ssymm_direct kernel for cblas_ssymm level 3 API 2025-09-30 15:05:33 +05:30
Chip Kerchner 92f09a6a98 Add BF16 sbgemm on RISCV. 2025-09-22 14:32:43 +00:00
Martin Kroeker cb6c4392a5 Make GEMM3M parameters available on 32bit X86-GENERIC 2025-09-10 22:44:14 +02:00
Martin Kroeker 202a7a0e2a Separate VORTEXM4 from VORTEX and ARMV9SME 2025-08-18 01:45:40 -07:00
Chris Sidebottom 114316f361 Optimize SBGEMM / BGEMM for NEOVERSEV1 further 
This changes the kernels to pack full SVE vectors and reduces the
overall complexity of the inner GEMM loop.
 2025-08-11 09:25:13 +00:00
Masato Nakagawa 7e29f11396 Multi-thread GEMM Performance Improvement on NeoverseV1 (DIVIDE_RATE=1) 2025-07-29 18:54:36 +09:00
Martin Kroeker c504aedca1 Merge pull request #5400 from Mousius/neoversev2-target 
Add NEOVERSEV2 target support
 2025-07-25 15:47:06 +02:00
Chris Sidebottom 87247daadc Add NEOVERSEV2 target support 
Did a quick run around to make `TARGET=NEVOERSEV2` build successfully.

Fixes #5385
 2025-07-24 12:40:31 +01:00
Chris Sidebottom ea2faf0c9a Add optimized BGEMM for NEOVERSEN2 target 
This re-uses the existing NEOVERSEN2 8x4 `sbgemm` kernel to implement `bgemm`.
 2025-07-24 10:59:28 +00:00
Chris SidebottomandYe Tao 740efd71c4 Add optimized BGEMM kernel for NEOVERSEV1 target 
This also improves the testing and generic kernel by re-using the BF16
conversion functions.

Built on top of https://github.com/OpenMathLib/OpenBLAS/pull/5357 and derived from https://github.com/OpenMathLib/OpenBLAS/pull/5287

Co-authored-by: Ye Tao <ye.tao@arm.com>
 2025-07-10 23:23:27 +00:00
Chris SidebottomandYe Tao f95e7b0e32 Add infrastructure for BGEMM 
Setting up all the infrastructure for BGEMM support in OpenBLAS, hopefully I found all the right places.

Derived mostly from the previous work done in https://github.com/OpenMathLib/OpenBLAS/pull/5287

Co-authored-by: Ye Tao <ye.tao@arm.com>
 2025-07-08 16:22:41 +01:00
Masato Nakagawa 5253c8f165 Multi-thread Performance Improvement of GEMM with DIVIDE_RATE=1 for 
A64FX.
 2025-06-30 21:35:16 +09:00
h-motoki bba75d5e45 GEMM_PREFERED_SIZE parameter has been changed for A64FX. 2025-06-27 19:37:36 +09:00
Martin Kroeker d96daa220d Merge pull request #5290 from Srangrang/develop 
Add support for FP16 to openBLAS and shgemm on RISCV
 2025-06-24 23:10:15 +02:00
davidz-ampere aa90ab4142 Add support for Ampere AmpereOne processors 2025-06-24 00:12:34 -04:00
davidz-ampere be68ef03b4 Add support for Ampere processors 2025-06-15 22:00:40 -04:00
gkdddd 670ec6f757 Added shgemm_kernel_8x8 for RISCV64_ZVL128B and shgemm_kernel_16x8 for RISCV64_ZVL256B 
Added HFLOAT16 support for RISCV64
Added shgemm_kernel_8x8 for RISCV64_ZVL128B and shgemm_kernel_16x8 for RISCV64_ZVL256B based on HFLOAT16
The instruction sets used are ZVFH and ZFH, which need to be supported by RVV1.0

Related to issue #5279
Co-authored-by Linjin Li <linjin_li@163.com>
 2025-06-03 20:14:30 +08:00
Srangrang 0a967797a1 Add FP16 support for RISCV 2025-05-27 14:34:57 +08:00
Martin Kroeker a34b487f22 Remove spurious cast from Alpha and Cell's DEFAULT_ALIGN 2025-04-09 17:25:46 +02:00
Vaisakh K V f66ca05b31 Merge branch 'develop' into topic/sgemm_direct_sme1 2025-02-13 14:54:37 +05:30
Vaisakh K V d23eb3b93e Support for SME1 based sgemm_direct kernel for cblas_sgemm level 3 API 
* Added ARMV9SME target
* Added SGEMM_DIRECT kernel based on SME1
 2025-02-13 14:51:21 +05:30
Ye Tao c748e6a338 optimized sbgemm kernel for neoverse-v1 (sve-256) 
Signed-off-by: Ye Tao <ye.tao@arm.com>
 2025-02-05 10:06:37 +00:00
Aditya Tewari 4379a6fbe3 * checkpoint sbgemm for SVE-256 2025-02-03 12:49:49 +00:00
Martin Kroeker 926e56e389 Align GEMM3M parameters for GENERIC with ZGEMM and add P/Q/R 2024-11-14 14:04:25 -08:00
Martin Kroeker a47b3c8867 Fix unroll parameter selection for MIPS64_GENERIC 2024-10-13 22:54:34 +02:00
Martin Kroeker 7c4f3638fd switch PPCG4 SGEMM kernel to 4x4 2024-10-03 22:00:15 +02:00
gxw 48698b2b1d LoongArch64: Rename core 
Use microarchitecture name instead of meaningless strings to name the core,
the legacy core is still retained.
1. Rename LOONGSONGENERIC to LA64_GENERIC
2. Rename LOONGSON3R5 to LA464
3. Rename LOONGSON2K1000 to LA264
 2024-09-29 09:35:21 +08:00
Chip Kerchner b1737698db Fix DEFAULTS in SBGEMM for POWER10. Also comparisons for SBGEMM unit test can be exactly due to epilison differences. 2024-08-13 07:01:21 -05:00
Piotr Kubaj 4c12090776 Fix build on FreeBSD/powerpc64* 2024-07-10 22:21:48 +00:00
gxw 6017ad7146 loongarch64: Update dgemm_kernel_16x4 to dgemm_kernel_16x6 2024-05-08 10:10:26 +08:00