76b4b8980f
Use -dumpversion with gcc only
2018-12-23 19:08:19 +01:00
49e0f485da
Add -mavx2 for TARGET=HASWELL if compiler supports and requires it
2018-12-23 17:26:09 +01:00
43c2b0eb55
Add -mavx2 to TARGET=HASWELL builds
...
to leverage improvements from PR#1921
2018-12-23 17:16:43 +01:00
942e229ed5
Merge pull request #1930 from martin-frbg/issue1908
...
Reflect ARMV8 target definition changes from PR1876
2018-12-23 15:06:33 +01:00
26a3402773
Reflect ARMV8 target definition changes from PR1876
...
and create config target directory for cross-compiles.
2018-12-23 12:26:01 +01:00
20033f992a
Merge pull request #1929 from martin-frbg/issue1924
...
Avoid taking the root of a negative number in simple threaded syrk
2018-12-23 09:03:58 +01:00
f343ed65b5
Avoid taking the root of a negative number
...
Fixes #1924 where numpy 1.17+ would report the (transient) FE_INVALID exception raised for the domain error.
2018-12-22 22:30:29 +01:00
a5a1118527
Merge pull request #1 from xianyi/develop
...
rebase
2018-12-22 22:13:44 +01:00
e23366e860
Merge pull request #1921 from fenrus75/haswelldgemm
...
Replicate some of the SKYLAKEX dgemm improvements also to HASWELL
2018-12-17 08:39:20 +01:00
b28f75cd7e
set GEMM_PREFERED_SIZE for HASWELL
...
Haswell likes a GEMM_PREFERED_SIZE of 16 to improve the split that the
threading code does to make it a nice multiple of the SIMD kernel size
2018-12-16 23:09:27 +00:00
d321448a63
dgemm: use dgemm_ncopy_8_skylakex.c also for Haswell
...
The dgemm_ncopy_8_skylakex.c code is not avx512 specific and gives
a nice performance boost for medium sized matrices
2018-12-16 23:09:22 +00:00
c43331ad0a
dgemm: Use the skylakex beta function also for haswell
...
it's more efficient for certain tall/skinny matrices
2018-12-16 23:09:17 +00:00
e8ca5a59a9
Merge pull request #1919 from fenrus75/haswelltuning
...
(sgemm) Apply some of the SKYLAKEX optimizations also to HASWELL
2018-12-16 20:11:05 +01:00
c4e23dd016
Update Makefile
2018-12-16 18:14:40 +01:00
cfc4acc221
typo
2018-12-16 16:19:51 +01:00
545c2b1bbb
Add -mavx2 on Haswell only if the compiler supports it
2018-12-16 13:09:19 +01:00
69d206440a
Make the skylakex/haswell sgemm code compile and run even with compilers without avx2 support
2018-12-16 00:19:41 +00:00
3843e3e017
use -maxv2 on haswell
2018-12-15 23:30:31 +01:00
fbcb14a74b
should be core-avx2
2018-12-15 20:18:59 +01:00
2a3190dc76
fix elseifeq and use older option core2-avx for compatibility
2018-12-15 20:17:44 +01:00
1ebe5c0f49
Add -march=haswell to HASWELL part of DYNAMIC_ARCH build
2018-12-15 19:35:35 +01:00
0586899a10
Use sgemm_ncopy_4_skylakex.c also for Haswell
...
sgemm_ncopy_4_skylakex.c uses SSE transpose operations where the
real perf win happens; this also works great for Haswell.
This gives double digit percentage gains on small and skinny matrices
2018-12-15 13:49:19 +00:00
00dc09ad19
Use the skylake sgemm beta code also for haswell
...
with a few small changes it's possible to use the skylake sgemm code
also for haswell, this gives a modest gain (10% range) for smallish
matrixes but does wonders for very skinny matrixes
2018-12-15 13:49:13 +00:00
78d877b54b
Merge pull request #1914 from fenrus75/smallmatrix
...
Add a "sgemm direct" mode for small matrixes
2018-12-13 19:08:14 +01:00
cdc668d82b
Add a "sgemm direct" mode for small matrixes
...
OpenBLAS has a fancy algorithm for copying the input data while laying
it out in a more CPU friendly memory layout.
This is great for large matrixes; the cost of the copy is easily
ammortized by the gains from the better memory layout.
But for small matrixes (on CPUs that can do efficient unaligned loads) this
copy can be a net loss.
This patch adds (for SKYLAKEX initially) a "sgemm direct" mode, that bypasses
the whole copy machinary for ALPHA=1/BETA=0/... standard arguments,
for small matrixes only.
What is small? For the non-threaded case this has been measured to be
in the M*N*K = 28 * 512 * 512 range, while in the threaded case it's
less, around M*N*K = 1 * 512 * 512
2018-12-13 13:47:31 +00:00
87718807f0
Merge pull request #1910 from martin-frbg/issue1909
...
Fix for DYNAMIC_ARCH builds made on a AVX512-capable host
2018-12-12 14:56:25 +01:00
51aec8e96b
make sure the added march=skylake-avx512 does not cause problems on Windows
2018-12-11 22:47:32 +01:00
06f7d78d70
Add -march=skylake-avx512 to SkylakeX part of DYNAMIC_ARCH builds
2018-12-11 21:10:38 +01:00
38cc638591
Avoid adding blanket march=skylake-avx512 to dynamic_arch builds
2018-12-11 21:09:26 +01:00
0bf6d74e5f
Fix typo in previous commit for arm dynamic arch
2018-12-07 19:37:33 +01:00
133c278ee5
Add DYNAMIC_CORE list for ARM64
...
cf #1908
2018-12-07 17:42:23 +01:00
2b355592e3
Make sure to use the arm version of dynamic.c in ARM64 DYNAMIC_ARCH
...
cf. #1908
2018-12-07 16:25:55 +01:00
ff3eb1d474
Merge pull request #1904 from martin-frbg/issue1870
...
Fix cmake parsing of GEMM kernels for ARMV8
2018-12-06 23:01:23 +01:00
0b09516678
Fix missing parameter in popen call
2018-12-06 18:33:05 +01:00
7639f2e1f0
Rewrite the conditional for OSX to fix cmake parsing on others
...
The Makefile variable parser in utils.cmake currently does not handle conditionals. Having the definitions for non-OSX last will at least make cmake builds work again on non-OSX platforms.
2018-12-06 14:04:27 +01:00
2fc712469d
Avoid creating spurious non-suffixed c/zgemm_kernels
...
Plain cgemm_kernel and zgemm_kernel are not used anywhere, only cgemm_kernel_b etc.
Needlessly building them (without any define like NN, CN, etc.) just happened to work on most platforms, but not on arm64. See #1870
2018-12-06 13:56:06 +01:00
6ba30e270d
Fix typo that broke CNRM2 on ARMV8 since 0.3.0
...
must have happened in my #1449
2018-12-06 13:42:25 +01:00
bf23518e36
Merge pull request #1903 from rengolin/armv8
...
Fix two mistakes on Arm64 builds
2018-12-05 22:10:53 +01:00
31a490ea88
Fix two mistakes on Arm64 builds
...
* Falkor is an ARMv8.0 with ARMv8.1 features, and chosing armv8.1-a for
march generates instructions it cannot cope with. Reverting it back
to armv8-a.
* ThunderX2's build was left with a #define VULCAN, which made it miss
the right compiler flags in Makefile.arm64, although it did create
the right library in the end.
2018-12-05 18:51:38 +00:00
701ea88347
Use p2align instead of align for OSX compatibility
...
fixes #1902
2018-12-03 13:06:43 +01:00
721c56c224
Merge pull request #1899 from brada4/fbsd12
...
Add mutually supported architecture mappings for FreeBSD12 ports
2018-12-03 12:50:27 +01:00
c5f8aeff2d
Merge branch 'develop' into fbsd12
2018-12-03 12:50:14 +01:00
8278cbe7f8
Merge pull request #1894 from pkubaj/patch-2
...
Use correct ARCH name on BSD powerpc64
2018-12-03 12:48:53 +01:00
ea6d1b96bd
Update Makefile.system
2018-12-03 08:59:10 +01:00
360374be62
Update with the changes from 0.3.4
2018-12-02 23:44:13 +01:00
f5acaad8f0
Increment version to 0.3.5.dev
2018-12-02 23:43:15 +01:00
93fa6b7b76
Increment version to 0.3.5.dev
2018-12-02 23:42:33 +01:00
c0827a7164
Update with changes from 0.3.4
2018-12-02 23:41:17 +01:00
86cff4effc
Merge pull request #1900 from xianyi/develop
...
Update from develop for 0.3.4
2018-12-02 23:40:21 +01:00
b028960aba
Merge branch 'release-0.3.0' into develop
2018-12-02 23:38:49 +01:00
Martin Kroeker
Arjan van de Ven
Renato Golin