19f17c8bc
Merge pull request #3893 from HellerZheng/develop by
2023-03-15 10:17:13 +0800
20511dfa6
Merge pull request #3919 from sergei-lewis/risc-v-latest-rvv-intrinsics by
2023-03-15 10:16:19 +0800
b83579d6a
(refs/pull/3939/head)
add riscv-v for nuclei UX900fd by
2023-03-13 11:41:01 +0800
8d6813ec4
Merge pull request #3938 from martin-frbg/issue3937 by
2023-03-10 15:33:07 +0100
19d679512
Merge pull request #3936 from martin-frbg/issue3933 by
2023-03-10 15:32:50 +0100
50c263716
(refs/pull/3938/head)
Correct DGEMM_UNROLL_M value for A53 in cross-compile by
2023-03-10 11:53:24 +0100
e298d613f
(refs/pull/3936/head)
initialize status variable for openblas_set_num_threads by
2023-03-08 23:43:15 +0100
05aa88268
add status variable for openblas_set_num_threads by
2023-03-08 23:41:57 +0100
9f9d0012a
observe thread limit imposed by openblas_set_num_threads() by
2023-03-08 23:38:34 +0100
fe5d3ca8e
Merge pull request #3935 from martin-frbg/omp_place_num by
2023-03-08 22:41:18 +0100
e38ab079a
(refs/pull/3935/head)
Fix OpenMP thread counting returning places rather than cores by
2023-03-08 19:17:33 +0100
606ef7b33
(refs/pull/3934/head)
Fix parallel make again by
2023-03-07 14:33:36 +0100
9feaaa3f3
Merge pull request #3932 from martin-frbg/issue3931 by
2023-03-03 12:01:25 +0100
8272dfc55
(refs/pull/3932/head)
Handle unrecognized ASM compiler (from Arm Compiler 22.1) by
2023-03-03 00:21:59 +0100
f616c8640
Merge pull request #3930 from sergei-lewis/dot-kernel-early-bail by
2023-03-02 16:46:25 +0100
cb0a70e0e
(refs/pull/3930/head)
dot.c early bail fix by
2023-03-02 09:51:10 +0000
8576476a9
(refs/pull/3928/head)
dot.c early bail fix by
2023-03-02 09:18:58 +0000
cda9cca89
(refs/pull/3927/head)
dot returns 0 for n==0, extend early bail check to cover this case by
2023-03-02 09:10:21 +0000
bdeee9e62
(refs/pull/3926/head)
put libtinfo5 install on separate instruction, remove ncurses by
2023-03-02 09:00:41 +0100
de63abeda
add ncurses to provide libtinfo for flang by
2023-03-02 08:46:05 +0100
9b61be454
(refs/pull/3919/head)
factoring riscv64/dot.c fix into separate PR as requested by
2023-03-01 17:40:42 +0000
5925178d0
Merge pull request #3924 from martin-frbg/numpy22025 by
2023-02-27 15:59:44 +0100
f58080278
Merge pull request #3923 from xctan/fix-cmake-riscv64 by
2023-02-27 09:39:30 +0800
3d27cbd9a
(refs/pull/3924/head)
avoid overflow in division by
2023-02-26 23:44:14 +0100
a39ced055
avoid overflow in division by
2023-02-26 23:42:20 +0100
6a0de3aa3
(refs/pull/3923/head)
Add missing RISC-V architecture in arch.cmake by
2023-02-26 20:21:57 +0800
240695862
* update intrinsics to match latest spec at https://github.com/riscv-non-isa/rvv-intrinsic-doc (in particular, __riscv_ prefixes for rvv intrinsics) * fix multiple numerical stability and corner case issues * add a script to generate arbitrary gemm kernel shapes * add a generic zvl256b target to demonstrate large gemm kernel unrolls by
2023-02-24 10:44:55 +0000
039e27545
Merge pull request #3915 from martin-frbg/issue3910 by
2023-02-24 07:41:33 +0100
38d6fb422
(refs/pull/3915/head)
Fix dependencies in builds with specified subsets of precision types by
2023-02-23 23:12:06 +0100
75d5e3eaf
Replace ifdefs and fix conditional definitions for including only selected precisions in DYNAMIC_ARCH by
2023-02-23 23:08:33 +0100
c0f341772
make SLARMM/DLARMM available to complex-only builds by
2023-02-22 00:38:30 +0100
e412bee31
fix GEMM kernel dependencies in builds that use only a subset of precisions by
2023-02-22 00:37:14 +0100
69256c2b6
fix GEMM kernel dependencies in builds for a subset of precisions by
2023-02-22 00:34:01 +0100
d80adf253
make SSYMV available to BUILD_DOUBLE-only builds by
2023-02-22 00:30:20 +0100
5481c328e
fix DYNAMIC_ARCH builds that use only a subset of precisions by
2023-02-22 00:28:25 +0100
ee4408282
fix DYNAMIC_ARCH builds that use only a subset of precisions by
2023-02-22 00:27:18 +0100
fa5ff7d19
slarmm/dlarmm are needed by COMPLEX/COMPLEX16-only builds too by
2023-02-22 00:25:12 +0100
cb76be5bd
Merge pull request #3914 from martin-frbg/lapack798 by
2023-02-19 19:18:18 +0100
1946eb4f4
(refs/pull/3914/head)
Fix bug in complex precision tests (c|z)het21 by
2023-02-19 10:30:16 +0100
10be02c89
Merge pull request #3909 from martin-frbg/lapack796 by
2023-02-15 12:56:47 +0100
85a03675f
(refs/pull/3909/head)
Fix ill-conditioned test matrix for DIAG=U in LIN testsuite test_rfp (LAPACK 678/796) by
2023-02-15 08:24:47 +0100
fa3bc574d
Merge pull request #3907 from martin-frbg/lapack794 by
2023-02-14 19:34:37 +0100
15c2571c9
Merge pull request #3906 from martin-frbg/lapack782 by
2023-02-14 19:34:15 +0100
24ceb0fc4
(refs/pull/3907/head)
Fix double subtraction of N_DEFLATE from istop in ?LAQZ0 (LAPACK 794) by
2023-02-14 12:43:41 +0100
f0f40a599
(refs/pull/3906/head)
Suppress warnings and delete unnecessary tests (LAPACK PR782) by
2023-02-14 12:06:21 +0100
2158dc64a
Merge pull request #3904 from martin-frbg/issue3901 by
2023-02-09 18:06:50 +0100
499c404f1
(refs/pull/3905/head)
Add linting workflow to GitHub Actions by
2023-02-09 10:09:56 +0000
ebe50458f
(refs/pull/3904/head)
Do not add a -tp to the flags of the nvc compiler if there is one already in CFLAGS by
2023-02-09 09:29:27 +0100
3dec11c66
Merge pull request #3902 from haampie/fix/parallel-build by
2023-02-08 15:52:29 +0100
bb7ae98df
(refs/pull/3902/head)
fix shared and tests prereqs by
2023-02-08 12:52:22 +0100
fdc1cdb10
(refs/pull/3754/head)
Merge pull request #3898 from martin-frbg/zen4fix by
2023-02-03 04:48:27 +0100
60dfba0d9
Merge pull request #3897 from martin-frbg/cortexx3-id by
2023-02-02 22:08:05 +0100
19a696f8f
(refs/pull/3898/head)
fix nested conditionals by
2023-02-02 19:59:49 +0100
e964ebd0d
Add compiler option for AVX512-capable Ryzen(4) by
2023-02-02 19:04:05 +0100
8e8651f0a
Supply necessary gcc option for AVX512-capable Ryzens by
2023-02-02 18:13:29 +0100
9ecfa9474
(refs/pull/3897/head)
Add part numbers for A715 and X3 aliased to A710/X2 by
2023-02-02 17:30:30 +0100
6876360a7
Merge pull request #3896 from antonio-rojas/patch-1 by
2023-02-02 17:24:36 +0100
ab3399d0c
Merge pull request #3895 from martin-frbg/issue3892 by
2023-02-02 15:45:45 +0100
21a18fdf2
(refs/pull/3896/head)
Fix USE_PERL option usage by
2023-02-02 13:12:41 +0100
afff3f551
(refs/pull/3895/head)
Fix typo that could clear EXTRALIB settings by
2023-02-02 10:43:44 +0100
7e97dddbf
Fix linking the library with libm on systems known to need it by
2023-02-02 10:43:01 +0100
98630db7a
Merge pull request #3894 from martin-frbg/homebrew-workflow2 by
2023-02-02 10:40:43 +0100
c718de14c
(refs/pull/3894/head)
Update nightly-Homebrew-build.yml by
2023-02-02 09:42:11 +0100
63cf4d016
(refs/pull/3893/head)
add riscv level3 C,Z kernel functions. by
2023-02-01 19:13:44 -0800
efe0b8424
(refs/pull/3891/head)
update for riscv V extension 1.0 and arbitrary shape gemm kernels by
2023-02-01 11:08:39 +0000
c19dff0a3
Fix T-Head RVV intrinsic API changes. by
2023-01-25 19:33:32 +0800
6c1076e13
Merge pull request #3886 from uniontech-lilinjie/develop by
2023-01-21 14:19:13 +0100
530da73b2
(refs/pull/3886/head)
fix typo by
2023-01-12 16:58:41 +0800
c4a3bd9f4
Merge pull request #3882 from martin-frbg/homebrew_workflow by
2023-01-04 18:45:47 +0100
5be8b43fa
(refs/pull/3882/head)
Disable dependents check to avoid python 3.11 conflict by
2023-01-04 16:56:38 +0100
2565e61d5
Merge pull request #3880 from martin-frbg/rework_msa by
2023-01-03 14:13:53 +0100
2a203eedd
(refs/pull/3880/head)
Make HAVE_MSA describe cpu capability, NO_MSA software/env by
2023-01-02 22:23:17 +0100
31fd13d04
MIPS: make HAVE_MSA reflect cpu capability and NO_MSA software/env by
2023-01-02 22:19:13 +0100
3f378cb6d
Move MIPS compiler options for MSA out of c_check by
2023-01-02 22:15:11 +0100
06b23cae0
(refs/pull/3878/head)
Update dynamic_arch.yml by
2023-01-01 12:53:53 +0100
716884441
Update dynamic_arch.yml by
2023-01-01 12:22:00 +0100
58363774e
Update Makefile.system by
2023-01-01 11:52:48 +0100
5d6fde58e
Merge pull request #3875 from martin-frbg/issue3874 by
2022-12-31 12:07:04 +0100
d4868babb
(refs/pull/3875/head)
Fix typos by
2022-12-29 23:07:55 +0100
3e64fa72c
Settings from Makefile(_kernel).conf should be available to DYNAMIC_ARCH kernel builds by
2022-12-29 23:05:22 +0100
274f0fd3e
(refs/pull/3876/head)
Always propagate HAVE_/NO_ values from ARCHCONFIG when FORCE is enabled by
2022-12-25 17:50:41 +0000
42cbfd2d1
Don't gate reading of Makefile.conf by
2022-12-25 15:49:06 +0000
18c99d3e6
Update dynamic_arm64.c by
2022-12-25 13:31:38 +0100
186a310f9
Update dynamic_arm64.c by
2022-12-25 12:22:48 +0100
ca3b5ae3f
Pass NO_SVE if set by
2022-12-25 12:19:20 +0100
f07f4db1c
On ARM64, check for SVE support by
2022-12-25 12:12:38 +0100
5ddcc5ffa
Update arch.cmake by
2022-12-25 10:37:39 +0100
18cac5a34
SVE-enabled ARM64 targets in DYNAMIC_ARCH require a recent compiler by
2022-12-25 10:25:13 +0100
d16261fbc
SVE-enabled targets in ARM64 DYNAMIC_ARCH require a recent compiler by
2022-12-25 10:19:02 +0100
5a9cd8779
Merge pull request #3868 from Mousius/sve-prefetch by
2022-12-24 10:52:29 +0100
efdd5b896
Merge pull request #3872 from martin-frbg/lapack775 by
2022-12-20 09:41:15 +0100
550f05215
Merge pull request #3873 from martin-frbg/workflow-flang by
2022-12-20 09:39:57 +0100
da028179e
(refs/pull/3873/head)
try libtinfo5 rather than libtinfo-devel by
2022-12-19 22:53:53 +0100
4d9d810b9
ensure libtinfo is available for Linux/flang by
2022-12-19 22:31:54 +0100
e436ca61b
(refs/pull/3872/head)
Fix uninitialized variables on quick return path (Reference-LAPACK PR775) by
2022-12-19 21:04:13 +0100
4c12fc5f5
Merge pull request #3852 from bartoldeman/sscal-opt by
2022-12-19 13:15:14 +0100
87d337c6c
Merge pull request #3867 from martin-frbg/issue3856 by
2022-12-19 13:14:20 +0100
7121e1668
(refs/pull/3869/head)
Use SVE kernel for SGEMM/DGEMM on Arm(R) Neoverse(TM) V1 by
2022-12-16 18:01:16 +0000
136122929
(refs/pull/3868/head)
Remove prefetches from SVE kernels by
2022-12-16 14:36:42 +0000
d5810b077
(refs/pull/3867/head)
Copy NO_WARMUP preset from Makefile.rule by
2022-12-16 09:16:49 +0100