platform/upstream/openblas.git
6 years agoMerge pull request #1302 from martin-frbg/nofortran-fix
Martin Kroeker [Thu, 14 Sep 2017 09:54:20 +0000 (11:54 +0200)]
Merge pull request #1302 from martin-frbg/nofortran-fix

Remove default FEXTRALIBS in NOFORTRAN case

6 years agoRemove default FEXTRALIBS in NOFORTRAN case
Martin Kroeker [Thu, 14 Sep 2017 07:21:04 +0000 (09:21 +0200)]
Remove default FEXTRALIBS in NOFORTRAN case

6 years agoMerge pull request #1288 from quickwritereader/develop
Martin Kroeker [Sat, 9 Sep 2017 21:47:17 +0000 (23:47 +0200)]
Merge pull request #1288 from quickwritereader/develop

Optimized standard Blas Level-1,2 (excluding nrm2 functions) for z13 (double precision). Issue 884

6 years agoMerge pull request #1293 from embray/cygwin/install
Martin Kroeker [Sat, 9 Sep 2017 21:46:27 +0000 (23:46 +0200)]
Merge pull request #1293 from embray/cygwin/install

More canonical installation on Cygwin

6 years agoMerge pull request #1299 from martin-frbg/race_fixes
Martin Kroeker [Sat, 9 Sep 2017 21:41:53 +0000 (23:41 +0200)]
Merge pull request #1299 from martin-frbg/race_fixes

Fix thread data races uncovered by gcc thread sanitizer

6 years agoConvert another caller of "allocation" to LOCK_COMMAND
Martin Kroeker [Sat, 9 Sep 2017 18:30:33 +0000 (20:30 +0200)]
Convert another caller of "allocation" to LOCK_COMMAND

... as the "allocation" code jumped to now does UNLOCK_COMMAND instead of blas_unlock

6 years agoFix thread data races
Martin Kroeker [Sat, 9 Sep 2017 17:07:06 +0000 (19:07 +0200)]
Fix thread data races

6 years agoFix thread data race in memory.c
Martin Kroeker [Sat, 9 Sep 2017 16:58:38 +0000 (18:58 +0200)]
Fix thread data race in memory.c

6 years agoMore canonical installation on Cygwin:
Erik M. Bray [Thu, 7 Sep 2017 12:18:56 +0000 (14:18 +0200)]
More canonical installation on Cygwin:

* The DLL is named cygopenblas.dll, not libopenblas.dll
* The import lib (still called libopenblas.dll.a) is installed

6 years agoOptimized standard Blas Level-1,2 (excluding nrm2 functions) for z13 (double precision)
Abdurrauf [Sat, 8 Apr 2017 17:51:15 +0000 (21:51 +0400)]
Optimized standard Blas Level-1,2 (excluding nrm2 functions) for z13 (double precision)

6 years agoMerge pull request #1290 from martin-frbg/imatcopy
Martin Kroeker [Sun, 3 Sep 2017 11:02:10 +0000 (13:02 +0200)]
Merge pull request #1290 from martin-frbg/imatcopy

Use in-place transform shortcut only if matrix is square

6 years agoUse in-place transform shortcut only if matrix is square
Martin Kroeker [Sun, 3 Sep 2017 07:52:55 +0000 (09:52 +0200)]
Use in-place transform shortcut only if matrix is square

6 years agoMerge pull request #1286 from martin-frbg/baytrail
Martin Kroeker [Sun, 27 Aug 2017 11:23:57 +0000 (13:23 +0200)]
Merge pull request #1286 from martin-frbg/baytrail

Fix coretype detection for Bay Trail Atom

6 years agoFix coretype detection for Bay Trail Atom
Martin Kroeker [Sun, 27 Aug 2017 11:06:54 +0000 (13:06 +0200)]
Fix coretype detection for Bay Trail Atom

My earlier PR #982 appears to have been incomplete in this regard - fixes #1285

6 years agoMerge pull request #1281 from sharkcz/armv8
Martin Kroeker [Sat, 19 Aug 2017 18:37:19 +0000 (20:37 +0200)]
Merge pull request #1281 from sharkcz/armv8

fix detection of generic ARMv8 CPUs

6 years agofix detection of generic ARMv8 CPUs
Dan Horák [Fri, 18 Aug 2017 12:53:29 +0000 (14:53 +0200)]
fix detection of generic ARMv8 CPUs

6 years agoMerge pull request #1277 from cconrads-scicomp/fix-installation-instructions
Martin Kroeker [Thu, 10 Aug 2017 21:42:23 +0000 (23:42 +0200)]
Merge pull request #1277 from cconrads-scicomp/fix-installation-instructions

Make: fix installation instructions

6 years agoMerge pull request #1276 from cconrads-scicomp/android_-lm_fix
Martin Kroeker [Thu, 10 Aug 2017 19:35:32 +0000 (21:35 +0200)]
Merge pull request #1276 from cconrads-scicomp/android_-lm_fix

ARM: do not add linker flag `-lm` unconditionally

6 years agoMerge pull request #1275 from cconrads-scicomp/recognize-gfortran-on-arm
Martin Kroeker [Thu, 10 Aug 2017 19:32:09 +0000 (21:32 +0200)]
Merge pull request #1275 from cconrads-scicomp/recognize-gfortran-on-arm

ARM: recognize gfortran pre-releases

6 years agoMake: show installation instructions after build
Christoph Conrads [Thu, 10 Aug 2017 18:22:26 +0000 (14:22 -0400)]
Make: show installation instructions after build

6 years agoMake: fix installation instructions
Christoph Conrads [Thu, 10 Aug 2017 16:47:18 +0000 (12:47 -0400)]
Make: fix installation instructions

The installation instructions as shown after successfully compiling
OpenBLAS are wrong because this arguments used during compilation have
to be provided to Make again.

6 years agoARM: do not add linker flag `-lm` unconditionally
Christoph Conrads [Thu, 10 Aug 2017 15:34:21 +0000 (11:34 -0400)]
ARM: do not add linker flag `-lm` unconditionally

On ARM the required math library depends on whether the soft floating
point ABI is used or not but this is already handled in
`Makefile.system`, lines 499-505.

6 years agoARM: recognize gfortran pre-releases
Christoph Conrads [Thu, 10 Aug 2017 15:48:29 +0000 (11:48 -0400)]
ARM: recognize gfortran pre-releases

Without proper recognition of gfortran versions such as
> GNU Fortran (GCC) 4.9.x 20150123 (prerelease)
OpenBLAS assumes the presence of the G77 compiler. Consequently,
`-lgfortran` is missing from the pkg-config file.

The aforementioned compiler is the gfortran compiler in the Android repo
in a commit tagged as `ndk-r14`, cf. Paul Mustière's gfortran build
instructions for Android at https://github.com/buffer51/android-gfortran

6 years agoMerge pull request #1264 from isuruf/dyn
Martin Kroeker [Tue, 8 Aug 2017 21:47:47 +0000 (23:47 +0200)]
Merge pull request #1264 from isuruf/dyn

Support DYNAMIC_ARCH with CMake

6 years agoMerge pull request #1268 from jirutka/travis-2
Martin Kroeker [Tue, 8 Aug 2017 19:54:35 +0000 (21:54 +0200)]
Merge pull request #1268 from jirutka/travis-2

Travis: Add jobs building with clang and disable job `LINUX64_MUSL USE_OPENMP=1`

6 years agoChange travis back to sudo true
Martin Kroeker [Tue, 8 Aug 2017 14:39:13 +0000 (16:39 +0200)]
Change travis back to sudo true

,,,to see if this has any influence on the recent ld SIGKILLS

6 years agoNo strncasecmp with MSVC
Isuru Fernando [Mon, 7 Aug 2017 18:37:25 +0000 (00:07 +0530)]
No strncasecmp with MSVC

6 years agoAdd commonobjs
Isuru Fernando [Mon, 7 Aug 2017 17:38:44 +0000 (23:08 +0530)]
Add commonobjs

6 years agoTest DYNAMIC_ARCH on appveyor
Isuru Fernando [Sun, 6 Aug 2017 13:47:31 +0000 (19:17 +0530)]
Test DYNAMIC_ARCH on appveyor

6 years agoMerge remote-tracking branch 'upstream/develop' into dyn
Isuru Fernando [Sun, 6 Aug 2017 13:37:00 +0000 (19:07 +0530)]
Merge remote-tracking branch 'upstream/develop' into dyn

6 years agoMerge pull request #1262 from martin-frbg/xmv_thread-splitting
Martin Kroeker [Sun, 6 Aug 2017 12:11:44 +0000 (14:11 +0200)]
Merge pull request #1262 from martin-frbg/xmv_thread-splitting

Make sure that range limit of last thread never exceeds data size

6 years agoMerge pull request #1256 from isuruf/develop
Martin Kroeker [Sun, 6 Aug 2017 12:10:18 +0000 (14:10 +0200)]
Merge pull request #1256 from isuruf/develop

Support compiling with clang on windows

6 years agoTravis: Add jobs building with clang
Jakub Jirutka [Sun, 6 Aug 2017 09:17:02 +0000 (11:17 +0200)]
Travis: Add jobs building with clang

6 years agoTravis: Disable job "LINUX64_MUSL USE_OPENMP=1"
Jakub Jirutka [Sun, 6 Aug 2017 09:06:03 +0000 (11:06 +0200)]
Travis: Disable job "LINUX64_MUSL USE_OPENMP=1"

https://github.com/xianyi/OpenBLAS/pull/1255#issuecomment-320494610

6 years agoBuild all branches so that appveyor works in forks
Isuru Fernando [Fri, 4 Aug 2017 02:34:16 +0000 (08:04 +0530)]
Build all branches so that appveyor works in forks

6 years agoNew utest for clang
Isuru Fernando [Fri, 4 Aug 2017 02:27:20 +0000 (07:57 +0530)]
New utest for clang

6 years agoMerge remote-tracking branch 'upstream/develop' into develop
Isuru Fernando [Fri, 4 Aug 2017 02:27:55 +0000 (07:57 +0530)]
Merge remote-tracking branch 'upstream/develop' into develop

6 years agoMerge pull request #1266 from ashwinyes/develop_thunderx2t99_fix_clang_compilation
Martin Kroeker [Thu, 3 Aug 2017 13:33:28 +0000 (15:33 +0200)]
Merge pull request #1266 from ashwinyes/develop_thunderx2t99_fix_clang_compilation

THUDNERX2T99: Fix clang compilation

6 years agoTHUDNERX2T99: Fix clang compilation
Ashwin Sekhar T K [Wed, 2 Aug 2017 18:28:45 +0000 (11:28 -0700)]
THUDNERX2T99: Fix clang compilation

6 years agoAdd missing EXCAVATOR
Isuru Fernando [Wed, 2 Aug 2017 13:09:04 +0000 (18:39 +0530)]
Add missing EXCAVATOR

6 years agoMerge pull request #1259 from isuruf/cmake
Martin Kroeker [Wed, 2 Aug 2017 13:31:05 +0000 (15:31 +0200)]
Merge pull request #1259 from isuruf/cmake

CMake Improvements

6 years agoFix extra whitespaces. CMake parser macro fails with it
Isuru Fernando [Wed, 2 Aug 2017 13:00:26 +0000 (18:30 +0530)]
Fix extra whitespaces. CMake parser macro fails with it

TODO: Fix the parser macro to strip trailing whitespaces

6 years agoAdd hemm3m and symm3m objects
Isuru Fernando [Wed, 2 Aug 2017 12:54:54 +0000 (18:24 +0530)]
Add hemm3m and symm3m objects

6 years agoFixes for dynamic_arch. almost there
Isuru Fernando [Wed, 2 Aug 2017 10:44:34 +0000 (16:14 +0530)]
Fixes for dynamic_arch. almost there

6 years agoUpdate trmv_thread.c
Martin Kroeker [Wed, 2 Aug 2017 10:03:54 +0000 (12:03 +0200)]
Update trmv_thread.c

6 years agoMerge pull request #1255 from jirutka/travis
Martin Kroeker [Wed, 2 Aug 2017 09:59:17 +0000 (11:59 +0200)]
Merge pull request #1255 from jirutka/travis

Travis: Rewrite config, build and test also on Alpine Linux (musl libc)

6 years agoMake sure that range_n of last thread never exceeds the actual data size when splitti...
Martin Kroeker [Tue, 1 Aug 2017 22:37:58 +0000 (00:37 +0200)]
Make sure that range_n of last thread never exceeds the actual data size when splitting the workload

6 years agoTravis: Allow job LINUX64_MUSL USE_OPENMP=1 to fail
Jakub Jirutka [Fri, 28 Jul 2017 16:08:44 +0000 (18:08 +0200)]
Travis: Allow job LINUX64_MUSL USE_OPENMP=1 to fail

See: https://github.com/xianyi/OpenBLAS/pull/1255#issuecomment-318692183

6 years agoTravis: Disable some gcc warnings to avoid exceeding Travis limit
Jakub Jirutka [Fri, 28 Jul 2017 12:32:17 +0000 (14:32 +0200)]
Travis: Disable some gcc warnings to avoid exceeding Travis limit

See: https://github.com/xianyi/OpenBLAS/pull/1255#issuecomment-318628666

6 years agoTravis: Build and test also on Alpine Linux (musl libc)
Jakub Jirutka [Fri, 28 Jul 2017 00:31:27 +0000 (02:31 +0200)]
Travis: Build and test also on Alpine Linux (musl libc)

Alpine jobs needs sudo (for chroot), so they run on VMs infrastructure.
That's why they are much slower than other jobs.

6 years agoTravis: Simplify configuration using Build Stages and APT addon
Jakub Jirutka [Fri, 28 Jul 2017 00:01:44 +0000 (02:01 +0200)]
Travis: Simplify configuration using Build Stages and APT addon

Using APT addon has nice side-effect - you don't need sudo anymore, so
it can run on Travis containers-based infrastructure that is much faster
than their VMs infrastructure (used when sudo is needed).

You've been still running on Ubuntu Presty builders, but new default is
Trusty. Thus I've explicitly set `dist: presty` to let it stay on
Presty, to not change build environment by this commit.

6 years agoMerge pull request #1260 from xianyi/revert-1254-xbmv_range
Martin Kroeker [Tue, 1 Aug 2017 18:07:32 +0000 (20:07 +0200)]
Merge pull request #1260 from xianyi/revert-1254-xbmv_range

Revert "Fix calculated range limit exceeding actual data size for last thread"

6 years agoconfigure kernel_core.h
Isuru Fernando [Tue, 1 Aug 2017 17:53:55 +0000 (23:23 +0530)]
configure kernel_core.h

6 years agoRevert "Fix calculated range limit exceeding actual data size for last thread"
Martin Kroeker [Tue, 1 Aug 2017 17:28:08 +0000 (19:28 +0200)]
Revert "Fix calculated range limit exceeding actual data size for last thread"

6 years agoconfigure setparam
Isuru Fernando [Tue, 1 Aug 2017 17:02:47 +0000 (22:32 +0530)]
configure setparam

6 years agoSupport DYNAMIC_ARCH with cmake
Isuru Fernando [Tue, 1 Aug 2017 16:01:55 +0000 (21:31 +0530)]
Support DYNAMIC_ARCH with cmake

6 years agoFix lapacke copying
Isuru Fernando [Tue, 1 Aug 2017 10:17:14 +0000 (15:47 +0530)]
Fix lapacke copying

6 years agoNo need of a temp file for f77blas.h
Isuru Fernando [Tue, 1 Aug 2017 09:57:19 +0000 (15:27 +0530)]
No need of a temp file for f77blas.h

6 years agoSupport out-of-source build
Isuru Fernando [Tue, 1 Aug 2017 09:40:41 +0000 (15:10 +0530)]
Support out-of-source build

6 years agoFix installing cblas.h and fix tabs
Isuru Fernando [Tue, 1 Aug 2017 09:28:49 +0000 (14:58 +0530)]
Fix installing cblas.h and fix tabs

6 years agoMerge pull request #1257 from martin-frbg/cgroups-prereq
Martin Kroeker [Tue, 1 Aug 2017 09:23:03 +0000 (11:23 +0200)]
Merge pull request #1257 from martin-frbg/cgroups-prereq

Rework __GLIBC_PREREQ checks to avoid breaking non-glibc builds

6 years agoDon't change timestamps
Isuru Fernando [Tue, 1 Aug 2017 05:32:00 +0000 (11:02 +0530)]
Don't change timestamps

6 years agoRework __GLIBC_PREREQ checks to avoid breaking non-glibc builds
Martin Kroeker [Mon, 31 Jul 2017 19:02:43 +0000 (21:02 +0200)]
Rework __GLIBC_PREREQ checks to avoid breaking non-glibc builds

6 years agoMerge pull request #1254 from martin-frbg/xbmv_range
Martin Kroeker [Mon, 31 Jul 2017 15:46:40 +0000 (17:46 +0200)]
Merge pull request #1254 from martin-frbg/xbmv_range

Fix calculated range limit exceeding actual data size for last thread

6 years agoRemove unnecessary line in appveyor
Isuru Fernando [Sat, 29 Jul 2017 18:30:37 +0000 (00:00 +0530)]
Remove unnecessary line in appveyor

6 years agoFix vcvarsall call in appveyor
Isuru Fernando [Sat, 29 Jul 2017 18:12:56 +0000 (23:42 +0530)]
Fix vcvarsall call in appveyor

6 years agoFix copying libopenblas.dll
Isuru Fernando [Sat, 29 Jul 2017 18:12:38 +0000 (23:42 +0530)]
Fix copying libopenblas.dll

6 years agoMake ARCH variable a CACHE variable
Isuru Fernando [Sat, 29 Jul 2017 18:00:15 +0000 (23:30 +0530)]
Make ARCH variable a CACHE variable

6 years agoTry adding RC to path
Isuru Fernando [Sat, 29 Jul 2017 16:46:53 +0000 (22:16 +0530)]
Try adding RC to path

6 years agovsvarsall in appveyor
Isuru Fernando [Sat, 29 Jul 2017 16:28:53 +0000 (21:58 +0530)]
vsvarsall in appveyor

6 years agoFix CMAKE_C_COMPILER in appveyor
Isuru Fernando [Sat, 29 Jul 2017 16:24:32 +0000 (21:54 +0530)]
Fix CMAKE_C_COMPILER in appveyor

6 years agoadd --yes to conda in appveyor.yml
Isuru Fernando [Sat, 29 Jul 2017 16:18:49 +0000 (21:48 +0530)]
add --yes to conda in appveyor.yml

6 years agobuild clang-cl first
Isuru Fernando [Sat, 29 Jul 2017 16:17:15 +0000 (21:47 +0530)]
build clang-cl first

6 years agoFix appveyor.yml
Isuru Fernando [Sat, 29 Jul 2017 16:07:48 +0000 (21:37 +0530)]
Fix appveyor.yml

6 years agoTest clang in appveyor.yml
Isuru Fernando [Sat, 29 Jul 2017 15:46:00 +0000 (21:16 +0530)]
Test clang in appveyor.yml

6 years agoNinja complains that file openblas.def does not exist
Isuru Fernando [Sat, 29 Jul 2017 15:30:32 +0000 (21:00 +0530)]
Ninja complains that file openblas.def does not exist

6 years agoclang on windows needs FU=''
Isuru Fernando [Sat, 29 Jul 2017 15:29:17 +0000 (20:59 +0530)]
clang on windows needs FU=''

6 years agotypedefs only for c
Isuru Fernando [Sat, 29 Jul 2017 15:08:16 +0000 (20:38 +0530)]
typedefs only for c

6 years agoFix complex support for MSVC headers
Isuru Fernando [Fri, 28 Jul 2017 06:20:29 +0000 (11:50 +0530)]
Fix complex support for MSVC headers

6 years agocheck compiler is msvc instead of msvc
Isuru Fernando [Fri, 28 Jul 2017 06:19:39 +0000 (11:49 +0530)]
check compiler is msvc instead of msvc

6 years agoFix range limit exceeding actual data size in last step
Martin Kroeker [Thu, 27 Jul 2017 22:27:02 +0000 (00:27 +0200)]
Fix range limit exceeding actual data size in last step

6 years agoFix range limit exceeding data size in last step
Martin Kroeker [Thu, 27 Jul 2017 22:21:53 +0000 (00:21 +0200)]
Fix range limit exceeding data size in last step

6 years agoFix range exceeding actual data size in quick_divide
Martin Kroeker [Thu, 27 Jul 2017 22:13:24 +0000 (00:13 +0200)]
Fix range exceeding actual data size in quick_divide

6 years agoMerge pull request #1249 from martin-frbg/cgroup
Martin Kroeker [Tue, 25 Jul 2017 21:31:57 +0000 (23:31 +0200)]
Merge pull request #1249 from martin-frbg/cgroup

Honor cgroup/cpuset limits when enumerating cpus

6 years agoHonor cgroup/cpuset limits when enumerating cpus
Martin Kroeker [Tue, 25 Jul 2017 20:47:34 +0000 (22:47 +0200)]
Honor cgroup/cpuset limits when enumerating cpus

6 years agoRevert "Honor cgroup/cpuset limits when enumerating cpus" (#1246)
Martin Kroeker [Mon, 24 Jul 2017 14:17:50 +0000 (16:17 +0200)]
Revert "Honor cgroup/cpuset limits when enumerating cpus" (#1246)

6 years agoMerge pull request #1236 from martin-frbg/l1cache
Zhang Xianyi [Mon, 24 Jul 2017 04:07:00 +0000 (12:07 +0800)]
Merge pull request #1236 from martin-frbg/l1cache

Use cpuid 4 with subleafs to query L1 cache size on Intel processors

6 years agoBump develop version for 0.3.0.
Zhang Xianyi [Mon, 24 Jul 2017 04:06:29 +0000 (12:06 +0800)]
Bump develop version for 0.3.0.

6 years agoMerge branch 'develop' upstream/0.2.20 v0.2.20
Zhang Xianyi [Mon, 24 Jul 2017 04:03:35 +0000 (12:03 +0800)]
Merge branch 'develop'

0.2.20 version

6 years agoUpdate doc for 0.2.20 version.
Zhang Xianyi [Mon, 24 Jul 2017 03:55:10 +0000 (11:55 +0800)]
Update doc for 0.2.20 version.

6 years agoMerge pull request #1239 from martin-frbg/cgroups
Zhang Xianyi [Mon, 24 Jul 2017 03:46:52 +0000 (11:46 +0800)]
Merge pull request #1239 from martin-frbg/cgroups

Honor cgroup/cpuset limits when enumerating cpus

6 years agoMerge pull request #1244 from martin-frbg/micmuc_cimatcopy
Zhang Xianyi [Mon, 24 Jul 2017 03:45:27 +0000 (11:45 +0800)]
Merge pull request #1244 from martin-frbg/micmuc_cimatcopy

Fix complex imatcopy for Trans cases with non-square matrix

6 years agoUse in-place transform shortcut only if matrix is square
Martin Kroeker [Fri, 21 Jul 2017 09:20:15 +0000 (11:20 +0200)]
Use in-place transform shortcut only if matrix is square

6 years agoAdd files via upload
Martin Kroeker [Thu, 20 Jul 2017 18:51:06 +0000 (20:51 +0200)]
Add files via upload

6 years agoExchange rows and cols in final omatcopy with BlasTrans
Martin Kroeker [Sat, 15 Jul 2017 20:02:53 +0000 (22:02 +0200)]
Exchange rows and cols in final omatcopy with BlasTrans

This is MicMuc's patch from #899

6 years agoMore fixes for silly misedits
Martin Kroeker [Sat, 15 Jul 2017 10:48:42 +0000 (12:48 +0200)]
More fixes for silly misedits

6 years agoFixup braces lost in previous edit
Martin Kroeker [Sat, 15 Jul 2017 09:53:28 +0000 (11:53 +0200)]
Fixup braces lost in previous edit

6 years agoMerge branch 'develop' into cgroups
Martin Kroeker [Sat, 15 Jul 2017 08:40:42 +0000 (10:40 +0200)]
Merge branch 'develop' into cgroups

6 years agoDisable ReLAPACK by default (#1238)
Martin Kroeker [Thu, 13 Jul 2017 20:01:47 +0000 (22:01 +0200)]
Disable ReLAPACK by default (#1238)

* Disable ReLAPACK by default; mention it in final build message if included

* Add files via upload

* Add files via upload

* Add files via upload

6 years agoMerge pull request #1214 from martin-frbg/relapack
Zhang Xianyi [Thu, 13 Jul 2017 12:31:08 +0000 (20:31 +0800)]
Merge pull request #1214 from martin-frbg/relapack

Initial import of ReLAPACK