Martin Kroeker [Sun, 11 Jul 2021 14:00:29 +0000 (16:00 +0200)]
Clean up some warnings
Martin Kroeker [Sun, 11 Jul 2021 05:20:19 +0000 (07:20 +0200)]
Merge pull request #3301 from martin-frbg/syr2bench
Handle OPENBLAS_LOOPS in SYR2 benchmark
Martin Kroeker [Sat, 10 Jul 2021 19:27:53 +0000 (21:27 +0200)]
Handle OPENBLAS_LOOPS in SYR2 benchmark
Martin Kroeker [Sat, 10 Jul 2021 18:48:53 +0000 (20:48 +0200)]
Merge pull request #3299 from martin-frbg/issue3298
Fix copy-paste error in LIBCORE assignment for Tiger Lake
Martin Kroeker [Sat, 10 Jul 2021 16:20:40 +0000 (18:20 +0200)]
Fix copy-paste error in LIBCORE assignment for Tiger Lake
Martin Kroeker [Thu, 8 Jul 2021 19:24:15 +0000 (21:24 +0200)]
Merge pull request #3296 from martin-frbg/issue3295
Support Zhaoxin/Centaur family 7 processors as Nehalem
Martin Kroeker [Thu, 8 Jul 2021 16:28:49 +0000 (18:28 +0200)]
Add vendor string Shanghai as successor to Centaur
Martin Kroeker [Thu, 8 Jul 2021 16:26:23 +0000 (18:26 +0200)]
Add vendor string Shanghai as the successor to Centaur
Martin Kroeker [Thu, 8 Jul 2021 10:23:15 +0000 (12:23 +0200)]
Recognize newer Zhaoxin/Centaur processors as Nehalem
Martin Kroeker [Thu, 8 Jul 2021 10:20:19 +0000 (12:20 +0200)]
Recognize newer Zhaoxin/Centaur cpus as Nehalem
Martin Kroeker [Wed, 7 Jul 2021 18:58:29 +0000 (20:58 +0200)]
Merge pull request #3265 from TAAPArthur/improve_portability
Removed use of non portable '-p' arg to install
Martin Kroeker [Wed, 7 Jul 2021 18:46:54 +0000 (20:46 +0200)]
Merge pull request #3293 from martin-frbg/issue3290
Enable (C)EXTRALIB as for any other platform when building the tests on RISCV C910V
Martin Kroeker [Wed, 7 Jul 2021 18:46:28 +0000 (20:46 +0200)]
Merge pull request #3292 from martin-frbg/syrk_limit
Add lower limit for multithreading in xSYRK
Martin Kroeker [Wed, 7 Jul 2021 18:45:57 +0000 (20:45 +0200)]
Merge pull request #3288 from martin-frbg/getrf-2
Add lower threshold for multithreading in ?GETRF
Martin Kroeker [Tue, 6 Jul 2021 18:21:07 +0000 (20:21 +0200)]
Remove code that disabled EXTRALIB on RISCV C910V
Martin Kroeker [Sun, 4 Jul 2021 15:00:26 +0000 (17:00 +0200)]
Add lower limit for multithreading
Martin Kroeker [Sun, 4 Jul 2021 14:59:43 +0000 (16:59 +0200)]
Handle OPENBLAS_LOOP
Martin Kroeker [Thu, 1 Jul 2021 22:19:06 +0000 (00:19 +0200)]
Merge pull request #3289 from martin-frbg/issue3283
Update README to mention availability of the Windows binaries in the Releases section
Martin Kroeker [Thu, 1 Jul 2021 17:24:35 +0000 (19:24 +0200)]
Mention availability of the Windows binaries in the Releases section
Martin Kroeker [Thu, 1 Jul 2021 15:41:05 +0000 (17:41 +0200)]
Add lower threshold for multithreading
Martin Kroeker [Thu, 1 Jul 2021 15:39:37 +0000 (17:39 +0200)]
Handle OPENBLAS_LOOPS for more stable results
Martin Kroeker [Thu, 1 Jul 2021 15:38:45 +0000 (17:38 +0200)]
Handle OPENBLAS_LOOPS and OPENBLAS_TEST options
Martin Kroeker [Wed, 30 Jun 2021 05:42:45 +0000 (07:42 +0200)]
Merge pull request #3284 from martin-frbg/potrf_potri
Add lower thresholds for multithreading in POTRF/POTRI and improve the related benchmark
Martin Kroeker [Tue, 29 Jun 2021 18:09:26 +0000 (20:09 +0200)]
Merge pull request #3287 from martin-frbg/appveyor-conda
Work around current conda/tqdm auto-update problem on Appveyor
Martin Kroeker [Tue, 29 Jun 2021 15:17:34 +0000 (17:17 +0200)]
Work around current conda/tqdm auto-update problem
Martin Kroeker [Sat, 26 Jun 2021 21:47:41 +0000 (23:47 +0200)]
Add lower threshold for multithreading in ?potrf and ?potri
Martin Kroeker [Sat, 26 Jun 2021 21:46:00 +0000 (23:46 +0200)]
Add OPENBLAS_LOOPS support to potrf/potrs/potri benchmark
Martin Kroeker [Sat, 26 Jun 2021 21:44:56 +0000 (23:44 +0200)]
Add interface declarations for ?potri
Martin Kroeker [Wed, 23 Jun 2021 11:05:17 +0000 (13:05 +0200)]
Merge pull request #3278 from brada4/A55
Add CORTEXA55 cpuid 0xd05 support
Martin Kroeker [Wed, 23 Jun 2021 08:11:01 +0000 (10:11 +0200)]
Update cpuid_arm64.c
User User-User [Tue, 22 Jun 2021 21:58:59 +0000 (23:58 +0200)]
OK
User User-User [Sun, 20 Jun 2021 14:40:43 +0000 (16:40 +0200)]
bugz
User User-User [Sun, 20 Jun 2021 13:32:42 +0000 (15:32 +0200)]
add to cmake listings too
User User-User [Sun, 20 Jun 2021 13:29:25 +0000 (15:29 +0200)]
remove misplaced file
User User-User [Sun, 20 Jun 2021 12:19:41 +0000 (14:19 +0200)]
dynamic_arch
User User-User [Sat, 19 Jun 2021 20:21:39 +0000 (22:21 +0200)]
add gcc8 version matching
User User-User [Sat, 19 Jun 2021 19:49:58 +0000 (21:49 +0200)]
copy conf
User User-User [Sat, 19 Jun 2021 19:37:51 +0000 (21:37 +0200)]
WiP CORTEX A55 support
Martin Kroeker [Wed, 16 Jun 2021 14:37:30 +0000 (16:37 +0200)]
Merge pull request #3276 from martin-frbg/issue3274
Add workaround for another macro name collision with Windows 10 SDK winnt.h
Martin Kroeker [Wed, 16 Jun 2021 11:41:38 +0000 (13:41 +0200)]
Merge pull request #3275 from martin-frbg/lapack580
Fix missing EXTERNAL declarations in LAPACK TESTING (LAPACK PR 580)
Martin Kroeker [Wed, 16 Jun 2021 11:41:16 +0000 (13:41 +0200)]
Fix typo
Martin Kroeker [Wed, 16 Jun 2021 10:32:34 +0000 (12:32 +0200)]
Work around another recent macro name collision with winnt.h
Martin Kroeker [Wed, 16 Jun 2021 10:17:25 +0000 (12:17 +0200)]
Modify defines for CR and RC to work around name collision on Windows
Martin Kroeker [Wed, 16 Jun 2021 07:43:39 +0000 (09:43 +0200)]
Declare DZASUM as EXTERNAL
Martin Kroeker [Wed, 16 Jun 2021 07:41:18 +0000 (09:41 +0200)]
Declare ZDROT as EXTERNAL
Martin Kroeker [Wed, 16 Jun 2021 07:40:15 +0000 (09:40 +0200)]
Declare SROT as EXTERNAL
Martin Kroeker [Wed, 16 Jun 2021 07:39:28 +0000 (09:39 +0200)]
Declare DROT as EXTERNAL
Martin Kroeker [Wed, 16 Jun 2021 07:35:12 +0000 (09:35 +0200)]
Declare CSROT as EXTERNAL
Martin Kroeker [Wed, 16 Jun 2021 07:33:23 +0000 (09:33 +0200)]
Declare SCASUM as EXTERNAL
Martin Kroeker [Tue, 15 Jun 2021 20:58:48 +0000 (22:58 +0200)]
Merge pull request #3273 from austinpagan/sbgemm_gcc10_fix
Power10: Fix for SBGEMM
Gordon Fossum [Tue, 15 Jun 2021 18:07:47 +0000 (13:07 -0500)]
Power10: Fix for SBGEMM
While testing bfloat16 sbgemm kernel, there are some failures for odd value inputs due to updating result for
additional bytes.
Martin Kroeker [Tue, 15 Jun 2021 14:14:20 +0000 (16:14 +0200)]
Merge pull request #3252 from martin-frbg/more_shortcuts
Further shortcuts for (small) cases that do not need buffer allocation
Martin Kroeker [Tue, 15 Jun 2021 12:50:14 +0000 (14:50 +0200)]
Merge pull request #3250 from martin-frbg/gemv-shortcut
Add shortcut for small-size S/D GEMV_N with increments of one
Martin Kroeker [Mon, 14 Jun 2021 11:00:33 +0000 (13:00 +0200)]
Merge pull request #3270 from ggouaillardet/topic/dznrm2_tx2
arm64: add the missing d9 register to the clobber list
Gilles Gouaillardet [Mon, 14 Jun 2021 08:01:28 +0000 (17:01 +0900)]
arm64: add the missing d9 register to the clobber list
Refs. numpy/numpy#18422
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
Martin Kroeker [Thu, 10 Jun 2021 16:05:47 +0000 (18:05 +0200)]
Merge pull request #3266 from martin-frbg/powerparam
Remove spurious casts from PPC parameters and fix compilation for older targets
Martin Kroeker [Thu, 10 Jun 2021 14:08:24 +0000 (16:08 +0200)]
Merge pull request #3260 from intelmy/sgemv_t_opt
Optimized sgemv_t for small N based on AVX512
Martin Kroeker [Thu, 10 Jun 2021 14:07:47 +0000 (16:07 +0200)]
Merge pull request #3264 from RajalakshmiSR/sbgemmp10
POWER10: Fixes for sbgemm kernel
Martin Kroeker [Thu, 10 Jun 2021 09:20:29 +0000 (11:20 +0200)]
Add prefetch values for power3
Martin Kroeker [Thu, 10 Jun 2021 09:19:40 +0000 (11:19 +0200)]
Add prefetch values for power3
Martin Kroeker [Thu, 10 Jun 2021 09:18:22 +0000 (11:18 +0200)]
Add prefetch values for power3
Martin Kroeker [Thu, 10 Jun 2021 09:17:33 +0000 (11:17 +0200)]
Add prefetch values for power3
Martin Kroeker [Thu, 10 Jun 2021 09:15:48 +0000 (11:15 +0200)]
Fix caxpy/zaxpy for big-endian
Martin Kroeker [Thu, 10 Jun 2021 09:14:03 +0000 (11:14 +0200)]
Fix inverted conditional for caxpy/zaxpy
Martin Kroeker [Thu, 10 Jun 2021 09:11:56 +0000 (11:11 +0200)]
fix c/zrot and sgemv for POWER5
Martin Kroeker [Thu, 10 Jun 2021 09:09:50 +0000 (11:09 +0200)]
Remove casts for PPC/POWER and complete parameters for POWER3/4
Arthur Williams [Thu, 10 Jun 2021 01:50:11 +0000 (20:50 -0500)]
Removed use of non portable '-p' arg to install
Not all versions of install support '-p' flag and it isn't worth failing
the build in the installed files' timestamps get updated.
Rajalakshmi Srinivasaraghavan [Wed, 9 Jun 2021 17:20:09 +0000 (12:20 -0500)]
POWER10: Fixes for sbgemm kernel
While testing bfloat16 sbgemm kernel, there are some failures
for odd value inputs due to array access beyond the boundary.
Ma, Yu [Tue, 8 Jun 2021 19:08:28 +0000 (15:08 -0400)]
Optimized sgemv_t for small N based on AVX512
Zhang Xianyi [Tue, 8 Jun 2021 08:26:56 +0000 (16:26 +0800)]
Merge pull request #3259 from zhaofengli/riscv64-fixes
riscv64 fixes
Zhaofeng Li [Mon, 7 Jun 2021 22:55:56 +0000 (22:55 +0000)]
riscv64: Add Makefile
Zhaofeng Li [Mon, 7 Jun 2021 22:50:23 +0000 (22:50 +0000)]
RISCV64_GENERIC: Use generic kernel for DSDOT for better precision
The implementation in `riscv64/dot.c` fails the `test_dsdot` test, and
the generic kernel seems to have better precision. Tested on SiFive
FU740 (HiFive Unmatched) and QEMU.
Also see #1469.
Zhaofeng Li [Mon, 7 Jun 2021 22:49:39 +0000 (22:49 +0000)]
riscv64/imin: Fix wrong comparison
Same as #1990.
Martin Kroeker [Sun, 6 Jun 2021 20:15:29 +0000 (22:15 +0200)]
Merge pull request #3258 from martin-frbg/hbaction
revert "try to work around gcc update problems" in Homebrew workflow
Martin Kroeker [Sun, 6 Jun 2021 17:17:36 +0000 (19:17 +0200)]
revert "try to work around gcc update problems"
...as homebrew has dropped at least gcc8 now
Martin Kroeker [Sat, 29 May 2021 20:28:00 +0000 (22:28 +0200)]
Add shortcuts for (small) cases that do not need expensive buffer allocation
Martin Kroeker [Sat, 29 May 2021 13:40:03 +0000 (15:40 +0200)]
revert symv changes for now
Martin Kroeker [Fri, 28 May 2021 07:38:48 +0000 (09:38 +0200)]
Fix copy-paste errors in variables used
Martin Kroeker [Thu, 27 May 2021 20:39:18 +0000 (22:39 +0200)]
Add shortcuts for (small) cases that do not need expensive buffer allocation
Martin Kroeker [Wed, 26 May 2021 20:02:34 +0000 (22:02 +0200)]
Add shortcut for small-size gemv_n with increments of one
Martin Kroeker [Wed, 26 May 2021 13:26:30 +0000 (15:26 +0200)]
Merge pull request #3249 from MikaelUrankar/develop
Fix typo
MikaelUrankar [Wed, 26 May 2021 10:14:57 +0000 (12:14 +0200)]
Fix typo
Martin Kroeker [Sat, 22 May 2021 20:38:09 +0000 (22:38 +0200)]
Merge pull request #3244 from martin-frbg/issue3237
Add fast path for small xSYR with INCX==1
Martin Kroeker [Sat, 22 May 2021 18:41:18 +0000 (20:41 +0200)]
Add fast path for small xSYR with INCX==1
Martin Kroeker [Sat, 22 May 2021 17:25:56 +0000 (19:25 +0200)]
Merge pull request #3243 from martin-frbg/lapack564
Fix spurious error exit test failures in the ?chktsqr tests (LAPACK564)
Martin Kroeker [Sat, 22 May 2021 17:25:28 +0000 (19:25 +0200)]
Merge pull request #3196 from guowangy/skylakex-gemm-batch-k
GEMM: skylake: improve the performance when m is small
Martin Kroeker [Sat, 22 May 2021 17:24:46 +0000 (19:24 +0200)]
Merge pull request #3242 from martin-frbg/issue3239
Handle inadvertent use of DYNAMIC_ARCH=0
Martin Kroeker [Sat, 22 May 2021 12:29:45 +0000 (14:29 +0200)]
Fix spurious error exit test failures in the ?chktsqr tests (LAPACK564)
Martin Kroeker [Sat, 22 May 2021 12:23:49 +0000 (14:23 +0200)]
Handle inadvertent use of DYNAMIC_ARCH=0
Martin Kroeker [Mon, 17 May 2021 15:49:01 +0000 (17:49 +0200)]
Merge pull request #3205 from intelmy/sgemv_n_opt
optimize on sgemv_n for small n
Martin Kroeker [Mon, 17 May 2021 15:32:23 +0000 (17:32 +0200)]
Merge pull request #3238 from martin-frbg/lapack555
Correct function name in error message from SLASQ2 (LAPACK PR555)
Martin Kroeker [Mon, 17 May 2021 12:47:14 +0000 (14:47 +0200)]
Correct function name in error message from SLASQ2 (Reference-LAPACK PR 555)
Martin Kroeker [Sun, 16 May 2021 15:17:18 +0000 (17:17 +0200)]
Merge pull request #3236 from martin-frbg/issue3234
Add -lm for FreeBSD on ARM/ARM64
Martin Kroeker [Sun, 16 May 2021 15:15:45 +0000 (17:15 +0200)]
Merge pull request #3235 from dnoan/develop
Update Makefile.arm64
Martin Kroeker [Sun, 16 May 2021 11:04:38 +0000 (13:04 +0200)]
Add -lm for FreeBSD on ARM/ARM64
Noan [Sun, 16 May 2021 09:49:13 +0000 (09:49 +0000)]
Update Makefile.arm64
Added -march and -mtune flags for EMAG processors when GCC 9 or later
Martin Kroeker [Sat, 15 May 2021 07:06:12 +0000 (09:06 +0200)]
Merge pull request #3228 from martin-frbg/issue3226
filter out -mavx flag on Sandybridge zgemm/ztrmm kernels
Martin Kroeker [Fri, 14 May 2021 23:04:09 +0000 (01:04 +0200)]
Merge pull request #3233 from martin-frbg/issue3230
Add autodetection for Intel Ice Lake SP
Martin Kroeker [Fri, 14 May 2021 21:28:45 +0000 (23:28 +0200)]
Merge pull request #3232 from martin-frbg/lapack553
Reduce stack size requirements in the LAPACK LIN tests (LAPACK PR 553)
Martin Kroeker [Fri, 14 May 2021 21:28:06 +0000 (23:28 +0200)]
Merge pull request #3231 from martin-frbg/issue3227
Support compilation with pre-C99 versions of MSVC