795285c58
(refs/pull/1934/head)
Fix thinko in skylake beta handling by
2018-12-24 18:49:50 +0000
d6818777d
Make sure that -fPIC is present if needed by
2018-12-23 23:47:37 +0100
5bd21ab6e
Make sure that -fPIC is present when needed by
2018-12-23 23:46:48 +0100
e1eab9650
Merge pull request #1931 from martin-frbg/pr1921 by
2018-12-23 23:15:54 +0100
76b4b8980
(refs/pull/1931/head)
Use -dumpversion with gcc only by
2018-12-23 19:08:19 +0100
49e0f485d
Add -mavx2 for TARGET=HASWELL if compiler supports and requires it by
2018-12-23 17:26:09 +0100
43c2b0eb5
Add -mavx2 to TARGET=HASWELL builds by
2018-12-23 17:16:43 +0100
942e229ed
Merge pull request #1930 from martin-frbg/issue1908 by
2018-12-23 15:06:33 +0100
26a340277
(refs/pull/1930/head)
Reflect ARMV8 target definition changes from PR1876 by
2018-12-23 12:26:01 +0100
20033f992
Merge pull request #1929 from martin-frbg/issue1924 by
2018-12-23 09:03:58 +0100
f343ed65b
(refs/pull/1929/head)
Avoid taking the root of a negative number by
2018-12-22 22:30:29 +0100
a5a111852
Merge pull request #1 from xianyi/develop by
2018-12-22 22:13:44 +0100
e23366e86
Merge pull request #1921 from fenrus75/haswelldgemm by
2018-12-17 08:39:20 +0100
b28f75cd7
(refs/pull/1921/head)
set GEMM_PREFERED_SIZE for HASWELL by
2018-12-16 23:08:31 +0000
d321448a6
dgemm: use dgemm_ncopy_8_skylakex.c also for Haswell by
2018-12-16 23:06:58 +0000
c43331ad0
dgemm: Use the skylakex beta function also for haswell by
2018-12-16 22:59:02 +0000
e8ca5a59a
Merge pull request #1919 from fenrus75/haswelltuning by
2018-12-16 20:11:05 +0100
c4e23dd01
(refs/pull/1919/head)
Update Makefile by
2018-12-16 18:14:40 +0100
cfc4acc22
typo by
2018-12-16 16:19:51 +0100
545c2b1bb
Add -mavx2 on Haswell only if the compiler supports it by
2018-12-16 13:09:19 +0100
69d206440
Make the skylakex/haswell sgemm code compile and run even with compilers without avx2 support by
2018-12-16 00:19:41 +0000
3843e3e01
use -maxv2 on haswell by
2018-12-15 23:30:31 +0100
fbcb14a74
should be core-avx2 by
2018-12-15 20:18:59 +0100
2a3190dc7
fix elseifeq and use older option core2-avx for compatibility by
2018-12-15 20:17:44 +0100
1ebe5c0f4
Add -march=haswell to HASWELL part of DYNAMIC_ARCH build by
2018-12-15 19:35:35 +0100
0586899a1
Use sgemm_ncopy_4_skylakex.c also for Haswell by
2018-12-15 13:43:07 +0000
00dc09ad1
Use the skylake sgemm beta code also for haswell by
2018-12-15 13:18:59 +0000
78d877b54
Merge pull request #1914 from fenrus75/smallmatrix by
2018-12-13 19:08:14 +0100
cdc668d82
(refs/pull/1914/head)
Add a "sgemm direct" mode for small matrixes by
2018-12-12 16:45:57 +0000
87718807f
Merge pull request #1910 from martin-frbg/issue1909 by
2018-12-12 14:56:25 +0100
51aec8e96
(refs/pull/1910/head)
make sure the added march=skylake-avx512 does not cause problems on Windows by
2018-12-11 22:47:32 +0100
91d6a5693
(refs/pull/1911/head)
reduce scope of avx512 cflags by
2018-12-11 22:33:36 +0100
863162a1c
init by
2018-12-11 22:32:23 +0100
06f7d78d7
Add -march=skylake-avx512 to SkylakeX part of DYNAMIC_ARCH builds by
2018-12-11 21:10:38 +0100
38cc63859
Avoid adding blanket march=skylake-avx512 to dynamic_arch builds by
2018-12-11 21:09:26 +0100
0bf6d74e5
Fix typo in previous commit for arm dynamic arch by
2018-12-07 19:37:33 +0100
133c278ee
Add DYNAMIC_CORE list for ARM64 by
2018-12-07 17:42:23 +0100
2b355592e
Make sure to use the arm version of dynamic.c in ARM64 DYNAMIC_ARCH by
2018-12-07 16:25:55 +0100
5f99b61aa
(refs/pull/1907/head)
try to maintain cache line alignment for odd BLAS L1 work splits by
2018-12-07 00:15:27 +0100
583973428
init by
2018-12-07 00:13:45 +0100
ff3eb1d47
Merge pull request #1904 from martin-frbg/issue1870 by
2018-12-06 23:01:23 +0100
0b0951667
Fix missing parameter in popen call by
2018-12-06 18:33:05 +0100
7639f2e1f
(refs/pull/1904/head)
Rewrite the conditional for OSX to fix cmake parsing on others by
2018-12-06 14:04:27 +0100
2fc712469
Avoid creating spurious non-suffixed c/zgemm_kernels by
2018-12-06 13:56:06 +0100
6ba30e270
Fix typo that broke CNRM2 on ARMV8 since 0.3.0 by
2018-12-06 13:42:25 +0100
bf23518e3
Merge pull request #1903 from rengolin/armv8 by
2018-12-05 22:10:53 +0100
31a490ea8
(refs/pull/1903/head)
Fix two mistakes on Arm64 builds by
2018-12-05 18:51:38 +0000
701ea8834
Use p2align instead of align for OSX compatibility by
2018-12-03 13:06:43 +0100
721c56c22
Merge pull request #1899 from brada4/fbsd12 by
2018-12-03 12:50:27 +0100
c5f8aeff2
(refs/pull/1899/head)
Merge branch 'develop' into fbsd12 by
2018-12-03 12:50:14 +0100
8278cbe7f
Merge pull request #1894 from pkubaj/patch-2 by
2018-12-03 12:48:53 +0100
ea6d1b96b
(refs/pull/1894/head)
Update Makefile.system by
2018-12-03 08:59:10 +0100
360374be6
Update with the changes from 0.3.4 by
2018-12-02 23:44:13 +0100
f5acaad8f
Increment version to 0.3.5.dev by
2018-12-02 23:43:15 +0100
93fa6b7b7
Increment version to 0.3.5.dev by
2018-12-02 23:42:33 +0100
c0827a716
(tag: v0.3.4)
Update with changes from 0.3.4 by
2018-12-02 23:41:17 +0100
86cff4eff
Merge pull request #1900 from xianyi/develop by
2018-12-02 23:40:21 +0100
b028960ab
(refs/pull/1900/head)
Merge branch 'release-0.3.0' into develop by
2018-12-02 23:38:49 +0100
3c9e3faed
fixup BSD naming of powerpc arch by
2018-12-02 23:24:53 +0100
44c81fd13
oops by
2018-12-02 20:27:53 +0100
26b371048
Add architecture mappings for FreeBSD12 by
2018-12-02 12:07:41 +0100
84e614d0f
init by
2018-12-02 12:05:15 +0100
dceff5542
Handle Android environments that identify as Linux (#1898) by
2018-12-01 20:56:11 +0100
191014f94
(refs/pull/1898/head)
Update system_check.cmake by
2018-12-01 18:53:53 +0100
c1a060645
Handle Android environments that identify as Linux by
2018-12-01 12:46:34 +0100
6c7b69108
Really revert xDOT changes from 1832 by
2018-11-30 21:32:01 +0100
5f4c550c2
Merge pull request #1892 from martin-frbg/mipsdot by
2018-11-30 21:28:21 +0100
731b2722b
Fix build on POWER, remove DragonFly, add NetBSD by
2018-11-30 16:04:07 +0000
f85ce54d4
Use correct Makefile on powerpc64 by
2018-11-30 16:05:49 +0000
bbeb8a0b2
(refs/pull/1893/head)
Fix build on POWER, remove DragonFly, add NetBSD by
2018-11-30 16:04:07 +0000
2601cd58a
remove surplus locking code , only enabled w x86, disabled or never enabled on all others by
2018-10-11 23:29:34 +0300
95a5542e3
(refs/pull/1892/head)
Revert DOT kernel changes from #1834 by
2018-11-30 11:16:24 +0100
7a2e1bc80
Use generic kernel for DSDOT/SDSDOT by
2018-11-30 10:57:09 +0100
35653e38b
Merge pull request #1834 from fengrl/develop by
2018-11-30 10:48:46 +0100
71e25ae42
Merge pull request #1890 from martin-frbg/issue1889 by
2018-11-29 15:47:35 +0100
97d729897
(refs/pull/1890/head)
call it OpenBLAS not just version by
2018-11-29 11:52:08 +0100
de0d0ed52
Improve formatting of config output by
2018-11-29 11:28:19 +0100
081ceb3e0
Propagate version number for openblas_get_config by
2018-11-29 00:12:04 +0100
a29ec458c
propagate verison number for openblas_config_version by
2018-11-29 00:10:49 +0100
816775e30
Add version information to openblas_get_config output by
2018-11-29 00:06:44 +0100
c067c1fb7
(refs/pull/1887/head)
_AXPY calibration for >2 cores by
2018-11-26 17:52:22 +0100
c6dff2803
init by
2018-11-26 17:49:25 +0100
b6363f453
Merge pull request #1885 from brada4/freebsd by
2018-11-25 22:20:13 +0100
19c4bdd8b
(refs/pull/1885/head)
Add return value so that freebsd system clang does not err out by
2018-11-25 21:35:01 +0100
f049a4c84
init by
2018-11-25 21:34:09 +0100
f72fdf525
Merge pull request #1875 from martin-frbg/issue1851 by
2018-11-25 20:53:46 +0100
5393759a9
Merge pull request #1869 from martin-frbg/axpy0 by
2018-11-25 20:52:49 +0100
5cf18e287
Merge pull request #1878 from kiwifb/PGI_f_check by
2018-11-25 20:51:50 +0100
910050985
Merge pull request #1876 from rengolin/armv8-cleanup by
2018-11-25 20:51:24 +0100
da5894a2e
(refs/pull/1884/head)
Limit _AXPY threading due to observed regression with excess threads by
2018-11-23 16:08:15 +0100
6c044d596
init by
2018-11-23 16:03:42 +0100
0184713e1
(refs/pull/1878/head)
Correct link flags for PGI compiler. by
2018-11-21 14:24:56 +1300
459cc8380
(refs/pull/1877/head)
initialize z/cdot result variable at declaration time to avoid confusing static checker by
2018-11-20 20:59:35 +0100
ae895f864
initialize result at declaration instead of quick-exit only by
2018-11-20 20:58:08 +0100
c1fa2c801
init by
2018-11-20 20:56:02 +0100
45c3c459e
Merge pull request #1868 from martin-frbg/aix_cpuid by
2018-11-20 17:25:57 +0100
113cb00b9
(refs/pull/1875/head)
fix missing parenthesis by
2018-11-19 21:01:36 +0100
519265170
Add CriticalSection handling instead of mutexes for Windows by
2018-11-19 17:58:22 +0100
310ea55f2
(refs/pull/1876/head)
Simplifying ARMv8 build parameters by
2018-11-16 15:45:12 +0000
2e6fae2aa
Serialize accesses to parallelized level3 functions from multiple callers by
2018-11-19 14:02:50 +0100