a83f01e0e
Merge pull request #1662 from martin-frbg/cmake-avx512 by
2018-07-03 17:40:09 +0200
a49203b48
(refs/pull/1663/head)
Double MAX_ALLOCATING_THREADS to fix segfaults with Go and Octave by
2018-07-03 17:35:54 +0200
ba586c3d1
(refs/pull/1661/head)
Ensure that the gotoblas lookup table is always initialized. by
2018-07-03 12:06:54 +0100
b74aef281
(refs/pull/1662/head)
Add -march=skylake-avx512 to AVX512 compile check and suppress its output by
2018-07-03 14:41:44 +0200
a9fa80500
Merge pull request #1660 from martin-frbg/issue1659 by
2018-07-02 17:48:19 +0200
9d15a3bd1
(refs/pull/1660/head)
Fix typo that broke compilation with DYNAMIC_ARCH and NO_AVX2 by
2018-07-02 14:40:41 +0200
c6aec89d1
(tag: v0.3.1)
Merge pull request #1657 from martin-frbg/release-0.3.0 by
2018-07-01 12:03:07 +0200
bbf212497
set version number to 0.3.2.dev by
2018-07-01 12:01:51 +0200
1392eba48
set version number to 0.3.2.dev by
2018-07-01 12:01:16 +0200
e6d771119
(refs/pull/1657/head)
remove dev suffix from version number by
2018-07-01 11:59:47 +0200
7a914347c
remove dev suffix from version number by
2018-07-01 11:58:57 +0200
61659f876
Merge pull request #1648 from martin-frbg/nofort by
2018-07-01 11:56:40 +0200
3a8f0a6a1
Merge pull request #1656 from xianyi/develop by
2018-07-01 11:55:21 +0200
3d3c19717
(refs/pull/1656/head)
Merge pull request #1655 from martin-frbg/issue1641 by
2018-07-01 08:41:22 +0200
24e344038
Merge pull request #1654 from martin-frbg/avx512check by
2018-07-01 01:17:03 +0200
4e9c34018
(refs/pull/1655/head)
Fix apparent off-by-one error in calculation of MAX_ALLOCATING_THREADS by
2018-06-30 23:57:50 +0200
f5243e8e1
(refs/pull/1654/head)
Add compiler option to avx512 test and hide test output by
2018-06-30 23:47:44 +0200
ba8388cee
Merge pull request #1651 from martin-frbg/avx512-nodgemm by
2018-06-30 17:48:03 +0200
6e54b0a02
(refs/pull/1651/head)
Disable the 16x2 DTRMM kernel on SkylakeX as well by
2018-06-30 17:31:06 +0200
40c8cbc3b
Merge pull request #1650 from martin-frbg/avx512-nodgemm by
2018-06-30 13:05:46 +0200
d3c9eb4c7
Merge pull request #1639 from martin-frbg/dyn_list by
2018-06-30 13:05:30 +0200
f0a8dc2ee
(refs/pull/1650/head)
Disable the AVX512 DGEMM kernel for now by
2018-06-30 11:34:48 +0200
cc92257ea
(refs/pull/1648/head)
Update Makefile by
2018-06-27 00:09:21 +0200
2aba1b165
Merge branch 'develop' into nofort by
2018-06-27 00:07:32 +0200
8396e9e77
Handle NOFORTRAN=0 by
2018-06-27 00:00:27 +0200
bfad307ed
Merge pull request #1647 from martin-frbg/armv7-dot by
2018-06-26 22:27:30 +0200
b83e4c60c
(refs/pull/1647/head)
Remove premature exit for INC_X or INC_Y zero by
2018-06-26 20:46:42 +0200
e344db269
Remove premature exit for INC_X or INC_Y zero by
2018-06-26 20:45:57 +0200
545b82efd
Remove premature exit for INC_X or INC_Y zero by
2018-06-26 20:45:00 +0200
e322a951f
Remove premature exit for INC_X or INC_Y zero by
2018-06-26 20:44:13 +0200
ff2f17103
Merge pull request #1644 from martin-frbg/revert-filterout by
2018-06-26 10:15:15 +0200
092175cfe
(refs/pull/1644/head)
Revert changes to NOFORTRAN handling from 952541e by
2018-06-26 08:09:52 +0200
750162a05
(refs/pull/1639/head)
Try gradual fallback for cores not in the dynamic core list by
2018-06-25 21:02:31 +0200
e6d93f20f
Merge pull request #2 from martin-frbg/develop by
2018-06-25 20:48:10 +0200
c38c65eb6
Merge pull request #1 from xianyi/develop by
2018-06-25 20:45:56 +0200
ce3651516
Merge pull request #1642 from oon3m0oo/develop by
2018-06-25 19:23:40 +0200
014406853
(refs/pull/1642/head)
Rewrite &= -> = and simplify the initial blocking phase. by
2018-06-25 13:53:11 +0100
1833a6707
Add support for a user-defined list of dynamic targets by
2018-06-23 19:42:15 +0200
0b2b83d9e
Add support for a user-defined list of dynamic targets by
2018-06-23 19:41:32 +0200
62cf769aa
Merge pull request #1638 from martin-frbg/issue1637 by
2018-06-23 15:01:02 +0200
eb71d61c7
(refs/pull/1638/head)
Expose CBLAS interface to BLAS extensions iXamin by
2018-06-23 13:31:09 +0200
9cf22b7d9
Build cblas_iXamin interfaces by
2018-06-23 13:27:30 +0200
cc66743b6
Merge pull request #1634 from oon3m0oo/develop by
2018-06-21 21:01:03 +0200
2aa0a5804
Use BLAS rather than CBLAS in test_fork.c (#1626) by
2018-06-21 17:47:45 +0100
28c28ed27
(refs/pull/1634/head)
Fix data races reported by TSAN. by
2018-06-21 11:13:57 +0100
f4f110732
(refs/pull/1626/head)
Use BLAS rather than CBLAS in test_fork.c by
2018-06-19 16:31:26 +0100
a399d0042
Further improvements to memory.c. (#1625) by
2018-06-20 21:04:03 +0100
f66b9c882
Merge pull request #1630 from martin-frbg/x86-march by
2018-06-20 21:51:57 +0200
2946c4602
Merge pull request #1631 from oon3m0oo/stack by
2018-06-20 21:51:38 +0200
05978528c
(refs/pull/1631/head)
Avoid declaring arrays of size 0 when making large stack allocations. by
2018-06-20 17:03:18 +0100
ef6f0b645
Merge pull request #1629 from martin-frbg/issue1628 by
2018-06-20 16:41:13 +0200
0c5b7b400
(refs/pull/1630/head)
Add -march=skylake-avx512 to flags if target is skylake x by
2018-06-20 15:16:19 +0200
4d7719ec3
(refs/pull/1625/head)
Further improvements to memory.c. by
2018-06-18 17:13:19 +0100
952541e84
(refs/pull/1629/head)
Need to use filter-out to handle NOFORTRAN not set by
2018-06-20 13:20:30 +0200
9369d3e6e
Modify NOFORTRAN tests to always check the value; fix rewriting of NO_FORTRAN by
2018-06-19 23:28:06 +0200
10b70c904
Handle erroneous user settings NOFORTRAN=0 and NO_FORTRAN by
2018-06-19 20:53:19 +0200
6a5ab083b
Handle special case of gfortran+clang+OpenMP by
2018-06-19 20:47:33 +0200
1f9e4f319
Handle special case of gfortran+clang+OpenMP by
2018-06-19 20:46:36 +0200
5a6a2bed9
Merge pull request #1623 from fenrus75/fast-thread by
2018-06-18 09:02:40 +0200
2d8cc7193
Support upcoming Intel Cannon Lake CPUs as Skylake X (#1621) by
2018-06-17 23:38:14 +0200
2ddc96c9e
(refs/pull/1623/head)
make WMB / MB safer on x86-64 by
2018-06-17 18:06:24 +0000
7e39ffe11
On x86-64, make MB/WMB compiler barriers by
2018-06-17 17:53:15 +0000
73de17664
Add missing barriers in gemm scheduler by
2018-06-17 17:50:43 +0000
6eb4b9ae7
Tune HASWELL SWITCH_RATIO as well by
2018-06-17 17:05:04 +0000
5c6f00836
Tune param.h for SkylakeX by
2018-06-17 15:47:50 +0000
d148ec4ea
Don't use _Atomic for jobs sometimes... by
2018-06-17 15:39:15 +0000
9e162146a
Only initialize the part of the jobs array that will get used by
2018-06-17 15:32:03 +0000
3242317ef
(refs/pull/1621/head)
Fix bad copypasta of code from cpuid_x86 by
2018-06-17 11:30:25 +0200
2369240a5
Support upcoming Cannon Lake as Skylake X by
2018-06-16 23:13:49 +0200
1dc140ce0
Support upcoming Cannon Lake as Skylake X by
2018-06-16 23:13:04 +0200
47bf0dba8
Add build-time option for OMP scheduler; document MULTITHREAD_THRESHOLD range (#1620) by
2018-06-15 11:25:05 +0200
b9d1ba2e3
(refs/pull/1620/head)
Amended description of GEMM_MULTITHREAD_THRESHOLD by
2018-06-15 09:20:17 +0200
12603b7db
Merge pull request #1618 from oon3m0oo/less_locking by
2018-06-15 00:10:29 +0200
a217049dc
Leave DEBUG unset by default by
2018-06-15 00:09:42 +0200
b6b4b3cfc
Allow choosing the OpenMP scheduler and add range hint for GEMM_MULTITHREAD_THRESHOLD by
2018-06-15 00:00:13 +0200
b218e1a3c
Allow changing the OpenMP scheduler by
2018-06-14 23:44:33 +0200
bf40f806e
(refs/pull/1618/head)
Remove the need for most locking in memory.c. by
2018-06-14 12:18:04 +0100
ed682a4a0
Merge pull request #1619 from martin-frbg/issue1580 by
2018-06-14 17:48:51 +0200
fcb77ab12
(refs/pull/1619/head)
Update OSX deployment target to 10.8 by
2018-06-14 16:57:58 +0200
26e1cfb65
Merge pull request #1607 from martin-frbg/dynarch by
2018-06-14 16:52:55 +0200
c628c6fa5
Merge pull request #1612 from oon3m0oo/cpus by
2018-06-14 16:51:31 +0200
67d81ab49
Merge pull request #1609 from martin-frbg/issue1529 by
2018-06-12 23:00:24 +0200
2f957947a
Merge pull request #1613 from xianyi/revert-1600-noyield by
2018-06-11 17:14:49 +0200
de8fff671
(refs/pull/1613/head, revert-1600-noyield)
Revert "Use usleep instead of sched_yield by default" by
2018-06-11 17:05:27 +0200
6f71c0fce
Return a somewhat sane default value for L2 cache size if cpuid retur… (#1611) by
2018-06-11 13:26:19 +0200
c2545b0fd
(refs/pull/1612/head)
Fixed a few more unnecessary calls to num_cpu_avail. by
2018-06-11 10:13:09 +0100
fc72f6868
(refs/pull/1611/head)
Update setparam-ref.c by
2018-06-11 09:11:13 +0200
6fa7e8ccd
Return a somewhat sane default value for L2 cache size if cpuid returned something unexpected by
2018-06-11 00:02:05 +0200
e65f45140
(refs/pull/1609/head)
include CMakePackageConfigHelpers by
2018-06-10 15:09:43 +0200
02634b549
Add template for OpenBLASConfig.cmake by
2018-06-10 09:25:46 +0200
0bea6bb9e
Create OpenBLASConfig.cmake from cmake as well by
2018-06-10 09:24:37 +0200
3313e4b94
Merge pull request #1608 from martin-frbg/issue874 by
2018-06-09 19:57:33 +0200
e9cd11768
(refs/pull/1608/head)
Enable parallel make on MS Windows by default by
2018-06-09 17:54:36 +0200
63f7395fb
(refs/pull/1607/head)
Move some DYNAMIC_ARCH targets to new DYNAMIC_OLDER option by
2018-06-09 16:31:38 +0200
1cbd8f3ae
Move some DYNAMIC_ARCH targets to new DYNAMIC_OLDER option by
2018-06-09 16:30:46 +0200
6c2d90ba7
Move some DYNAMIC_ARCH targets to new DYNAMIC_OLDER option by
2018-06-09 16:29:17 +0200
0297b3211
Merge pull request #1605 from oon3m0oo/develop by
2018-06-09 12:42:34 +0200
66316b9f4
(refs/pull/1605/head)
Improve performance of GEMM for small matrices when SMP is defined. by
2018-06-07 14:54:42 +0100
e09fdbaa0
(refs/pull/1604/merge)
Merge 4b0de7690d into 6adc4b7b36 by
2018-06-07 13:57:52 +0000
4b0de7690
(refs/pull/1604/head)
Improve performance of GEMM for small matrices when SMP is defined. by
2018-06-07 14:54:42 +0100