platform/upstream/openblas.git
5 years agoMerge pull request #1858 from brada4/buff-1847
Martin Kroeker [Wed, 7 Nov 2018 07:46:55 +0000 (08:46 +0100)]
Merge pull request #1858 from brada4/buff-1847

Add minimum threshold for number of buffers

5 years agolets fit it in one 4k page
Andrew [Tue, 6 Nov 2018 17:51:24 +0000 (17:51 +0000)]
lets fit it in one 4k page

5 years agoadd low bound for number of buffers
Andrew [Tue, 6 Nov 2018 09:40:13 +0000 (09:40 +0000)]
add low bound for number of buffers

5 years agoinit
Andrew [Tue, 6 Nov 2018 09:39:05 +0000 (09:39 +0000)]
init

5 years agoMerge pull request #1846 from fenrus75/threadsize
Martin Kroeker [Fri, 2 Nov 2018 12:18:01 +0000 (13:18 +0100)]
Merge pull request #1846 from fenrus75/threadsize

gemm/dgemm: add a way for an arch kernel to specify preferred sizes

5 years agoMerge pull request #1850 from martin-frbg/issue1811
Martin Kroeker [Fri, 2 Nov 2018 08:50:51 +0000 (09:50 +0100)]
Merge pull request #1850 from martin-frbg/issue1811

Restore Android/ARMv7 build fix from #778

5 years agoMerge pull request #1849 from martin-frbg/aix_install2
Martin Kroeker [Thu, 1 Nov 2018 19:39:16 +0000 (20:39 +0100)]
Merge pull request #1849 from martin-frbg/aix_install2

Use installbsd on AIX

5 years agoRestore Android/ARMv7 build fix from #778
Martin Kroeker [Thu, 1 Nov 2018 17:50:25 +0000 (18:50 +0100)]
Restore Android/ARMv7 build fix from #778

for #1811

5 years agoUse installbsd on AIX
Martin Kroeker [Thu, 1 Nov 2018 17:26:08 +0000 (18:26 +0100)]
Use installbsd on AIX

(and fix misplaced parenthesis from previous commit). See #1803

5 years agoMerge pull request #1845 from martin-frbg/aix_install
Martin Kroeker [Thu, 1 Nov 2018 08:53:10 +0000 (09:53 +0100)]
Merge pull request #1845 from martin-frbg/aix_install

Accomodate AIX install, which has different syntax

5 years agosgemm/dgemm: add a way for an arch kernel to specify prefered sizes
Arjan van de Ven [Thu, 1 Nov 2018 01:43:20 +0000 (01:43 +0000)]
sgemm/dgemm: add a way for an arch kernel to specify prefered sizes

The current gemm threading code can make very unfortunate choices, for
example on my 10 core system a 1024x1024x1024 matrix multiply ends up
chunking into blocks of 102... which is not a vector friendly size
and performance ends up horrible.

this patch adds a helper define where an architecture can specify
a preference for size multiples.
This is different from existing defines that are minimum sizes and such.

The performance increase with this patch for the 1024x1024x1024 sgemm
is 2.3x (!!)

5 years agoskylakex: Make the sgemm/dgemm beta code robust for a N=0 or M=0 case
Arjan van de Ven [Thu, 1 Nov 2018 01:42:09 +0000 (01:42 +0000)]
skylakex: Make the sgemm/dgemm beta code robust for a N=0 or M=0 case

in the threading code there are cases where N or M can become 0,
and the optimized beta code did not handle this well, leading
to a crash

during the audit for the crash a few edge conditions on the if statements
were found and fixed as well

5 years agoAccomodate AIX install, which has different syntax
Martin Kroeker [Wed, 31 Oct 2018 20:50:34 +0000 (21:50 +0100)]
Accomodate AIX install, which has different syntax

for #1803

5 years agoMerge pull request #1843 from martin-frbg/aix_numprocs
Martin Kroeker [Wed, 31 Oct 2018 20:25:15 +0000 (21:25 +0100)]
Merge pull request #1843 from martin-frbg/aix_numprocs

Add get_num_procs implementation for AIX

5 years agoAdd get_num_procs implementation for AIX
Martin Kroeker [Wed, 31 Oct 2018 17:38:22 +0000 (18:38 +0100)]
Add get_num_procs implementation for AIX

(and copy HAIKU implementation to the non-TLS version of the code as well)

5 years agoMerge pull request #1837 from embray/set-num-thread-after-fork
Martin Kroeker [Tue, 30 Oct 2018 11:41:24 +0000 (12:41 +0100)]
Merge pull request #1837 from embray/set-num-thread-after-fork

Ensure that blas_thread_init has been called in openblas_set_num_threads

5 years agoensure that threading has been initialized in the first place before calling openblas...
Erik M. Bray [Sun, 28 Oct 2018 21:16:52 +0000 (21:16 +0000)]
ensure that threading has been initialized in the first place before calling openblas_set_num_threads

5 years agoMerge pull request #1836 from martin-frbg/zen2core
Martin Kroeker [Sun, 28 Oct 2018 19:00:01 +0000 (20:00 +0100)]
Merge pull request #1836 from martin-frbg/zen2core

Fix detection of Ryzen2 (missing CORE_ZEN)

5 years agoFix detection of Ryzen2 (missing CORE_ZEN)
Martin Kroeker [Sun, 28 Oct 2018 17:36:55 +0000 (18:36 +0100)]
Fix detection of Ryzen2 (missing CORE_ZEN)

5 years agoMerge pull request #1829 from ashwinyes/develop_aarch64_dynamic_arch_support
Martin Kroeker [Tue, 23 Oct 2018 16:14:28 +0000 (18:14 +0200)]
Merge pull request #1829 from ashwinyes/develop_aarch64_dynamic_arch_support

Add DYNAMIC_ARCH support for ARM64

5 years agoARM64: Enable DYNAMIC_ARCH
Ashwin Sekhar T K [Thu, 18 Oct 2018 12:15:45 +0000 (05:15 -0700)]
ARM64: Enable DYNAMIC_ARCH

Enable DYNAMIC_ARCH feature on ARM64. This patch uses the cpuid
feature in linux kernel to detect the core type at runtime
(https://www.kernel.org/doc/Documentation/arm64/cpu-feature-registers.txt).

If this feature is missing in kernel, then the user should use the
OPENBLAS_CORETYPE env variable to select the desired core type.

5 years agoARM64: Remove #define ARMV8 for THUNDERX
Ashwin Sekhar T K [Mon, 22 Oct 2018 08:49:16 +0000 (01:49 -0700)]
ARM64: Remove #define ARMV8 for THUNDERX

5 years agoARM64: Fix DYNAMIC_ARCH compilation for cores which dont use GEMM3M
Ashwin Sekhar T K [Thu, 18 Oct 2018 12:13:02 +0000 (05:13 -0700)]
ARM64: Fix DYNAMIC_ARCH compilation for cores which dont use GEMM3M

5 years agoARM64: Move parameters from parameter.c to param.h
Ashwin Sekhar T K [Thu, 18 Oct 2018 12:02:23 +0000 (05:02 -0700)]
ARM64: Move parameters from parameter.c to param.h

Remove the runtime setting of P, Q, R parameters for
targets ARMV8, THUNDERX2T99. Instead set them as constants
in param.h at compile time.

5 years agoARM64: Remove XGENE1 references
Ashwin Sekhar T K [Thu, 18 Oct 2018 11:51:24 +0000 (04:51 -0700)]
ARM64: Remove XGENE1 references

Remove XGENE1 target as the implementation for the
same is incomplete. Moreover whoever wishes to use
on XGENE1 can use the generic ARMV8 target as there
are no XGENE1 specific optimizations in OpenBLAS.

5 years agoMerge pull request #1825 from brada4/hemv
Martin Kroeker [Sun, 21 Oct 2018 18:34:05 +0000 (20:34 +0200)]
Merge pull request #1825 from brada4/hemv

Delay _hemv threading in attempt to address #1820

5 years agore-arrange new code for readability
Andrew [Sat, 20 Oct 2018 18:37:53 +0000 (21:37 +0300)]
re-arrange new code for readability

5 years agoAttempt to tame _hemv threading #1820
Andrew [Sat, 20 Oct 2018 08:13:29 +0000 (11:13 +0300)]
Attempt to tame _hemv threading #1820

5 years agoinit
Andrew [Sat, 20 Oct 2018 07:55:04 +0000 (10:55 +0300)]
init

5 years agoMerge pull request #1821 from ashwinyes/develop_aarch64_armv8neonkernels
Martin Kroeker [Thu, 18 Oct 2018 06:13:05 +0000 (08:13 +0200)]
Merge pull request #1821 from ashwinyes/develop_aarch64_armv8neonkernels

Use ThunderX2 Neon Kernels for ARMV8 Target

5 years agoARM64: Use THUNDERX2T99 Neon Kernels for ARMV8
Ashwin Sekhar T K [Wed, 17 Oct 2018 15:11:27 +0000 (08:11 -0700)]
ARM64: Use THUNDERX2T99 Neon Kernels for ARMV8

Currently the generic ARMV8 target uses C implementations
for many routines. Replace these with the neon implementations
written for THUNDERX2T99 target which are upto 6x faster for
certain routines.

5 years agoARM64: Remove dependency of THUNDERX2T99 Makefile on CORTEXA57 Makefile
Ashwin Sekhar T K [Wed, 17 Oct 2018 15:02:40 +0000 (08:02 -0700)]
ARM64: Remove dependency of THUNDERX2T99 Makefile on CORTEXA57 Makefile

5 years agoARM64: Remove dependency of THUNDERX Makefile on ARMV8 Makefile
Ashwin Sekhar T K [Wed, 17 Oct 2018 15:02:16 +0000 (08:02 -0700)]
ARM64: Remove dependency of THUNDERX Makefile on ARMV8 Makefile

5 years agoARM64: Remove dependency of CORTEXA57 Makefile on ARMV8 Makefile
Ashwin Sekhar T K [Wed, 17 Oct 2018 15:01:45 +0000 (08:01 -0700)]
ARM64: Remove dependency of CORTEXA57 Makefile on ARMV8 Makefile

5 years agoARM64: Remove dependency of XGENE1 Makefile on ARMV8 Makefile
Ashwin Sekhar T K [Wed, 17 Oct 2018 15:01:27 +0000 (08:01 -0700)]
ARM64: Remove dependency of XGENE1 Makefile on ARMV8 Makefile

5 years agoMerge pull request #1815 from fenrus75/sgemm_beta_fix
Martin Kroeker [Sun, 14 Oct 2018 17:57:34 +0000 (19:57 +0200)]
Merge pull request #1815 from fenrus75/sgemm_beta_fix

enable the SGEMM/SKX C based kernel

5 years agoenable the SGEMM/SKX C based kernel
Arjan van de Ven [Fri, 12 Oct 2018 09:30:35 +0000 (09:30 +0000)]
enable the SGEMM/SKX C based kernel

In QA the final bug was found so now the sklyakex sgemm C based kernel can
be activated....

5 years agoMerge pull request #1812 from martin-frbg/issue1806-2
Martin Kroeker [Thu, 11 Oct 2018 19:51:31 +0000 (21:51 +0200)]
Merge pull request #1812 from martin-frbg/issue1806-2

Use KERNEL_DEFINITIONS rather than COMMON_OPTS to pass -march=skylake…

5 years agoUse KERNEL_DEFINITIONS rather than COMMON_OPTS to pass -march=skylake-avx512
Martin Kroeker [Thu, 11 Oct 2018 09:03:27 +0000 (11:03 +0200)]
Use KERNEL_DEFINITIONS rather than COMMON_OPTS to pass -march=skylake-avx512

5 years agoMerge pull request #1808 from martin-frbg/issue1806
Martin Kroeker [Thu, 11 Oct 2018 05:48:08 +0000 (07:48 +0200)]
Merge pull request #1808 from martin-frbg/issue1806

Add -march=skylake-avx512 to CFLAGS when the target is Skylake

5 years agoMerge pull request #1807 from xianyi/revert-1798-cmake-avx512
Martin Kroeker [Thu, 11 Oct 2018 05:47:53 +0000 (07:47 +0200)]
Merge pull request #1807 from xianyi/revert-1798-cmake-avx512

Revert "Add -march=skylake-avx512 when required"

5 years agoSyntax fix
Martin Kroeker [Wed, 10 Oct 2018 21:47:35 +0000 (23:47 +0200)]
Syntax fix

5 years agoAdd -march=skylake-avx512 to CFLAGS when the target is Skylake
Martin Kroeker [Wed, 10 Oct 2018 17:22:01 +0000 (19:22 +0200)]
Add -march=skylake-avx512 to CFLAGS when the target is Skylake

Should fix 1806 and #1801

5 years agoRevert "Add -march=skylake-avx512 when required"
Martin Kroeker [Wed, 10 Oct 2018 17:15:32 +0000 (19:15 +0200)]
Revert "Add -march=skylake-avx512 when required"

5 years agoMerge pull request #1802 from martin-frbg/issue1801
Martin Kroeker [Wed, 10 Oct 2018 06:52:53 +0000 (08:52 +0200)]
Merge pull request #1802 from martin-frbg/issue1801

Use avx512 workaround with msys2/mingw64 as well

5 years agoMerge pull request #1804 from fenrus75/sgemm
Martin Kroeker [Wed, 10 Oct 2018 06:50:44 +0000 (08:50 +0200)]
Merge pull request #1804 from fenrus75/sgemm

Add a C+intrinsics version of the SGEMM/skylakex kernel

5 years agoAdd a C+intrinsics version of the SGEMM/skylakex kernel
Arjan van de Ven [Wed, 10 Oct 2018 01:49:22 +0000 (01:49 +0000)]
Add a C+intrinsics version of the SGEMM/skylakex kernel

for most sizes this is 1.2x to 1.4x faster than the current code

5 years agoMerge pull request #1800 from fengrl/patch-1
Martin Kroeker [Tue, 9 Oct 2018 08:56:37 +0000 (10:56 +0200)]
Merge pull request #1800 from fengrl/patch-1

Update common_mips64.h for the 1st loop of blas_memory_alloc

5 years agoMerge pull request #1792 from martin-frbg/cmakesuffix
Martin Kroeker [Tue, 9 Oct 2018 08:34:52 +0000 (10:34 +0200)]
Merge pull request #1792 from martin-frbg/cmakesuffix

Improve CMake help output and add SYMBOLPREFIX and -SUFFIX options

5 years agoUse cygwin compilation workaround for avx512 on msys2/mingw64 as well
Martin Kroeker [Tue, 9 Oct 2018 08:31:59 +0000 (10:31 +0200)]
Use cygwin compilation workaround for avx512 on msys2/mingw64 as well

5 years agoMerge pull request #1799 from martin-frbg/issue1796
Martin Kroeker [Tue, 9 Oct 2018 06:20:52 +0000 (08:20 +0200)]
Merge pull request #1799 from martin-frbg/issue1796

Handle conflicting usage of ARCH in at least some BSD environments

5 years agoMerge pull request #1793 from fenrus75/ncopy
Martin Kroeker [Tue, 9 Oct 2018 06:19:14 +0000 (08:19 +0200)]
Merge pull request #1793 from fenrus75/ncopy

Add optimized *copy versions for skylakex

5 years agoUpdate common_mips64.h
fengrl [Tue, 9 Oct 2018 03:20:16 +0000 (11:20 +0800)]
Update common_mips64.h

5 years agoCatch conflicting usage of ARCH in at least some BSD environments
Martin Kroeker [Mon, 8 Oct 2018 20:29:35 +0000 (22:29 +0200)]
Catch conflicting usage of ARCH in at least some BSD environments

fixes #1796

5 years agoUse override for ARCH in make.inc
Martin Kroeker [Mon, 8 Oct 2018 20:26:59 +0000 (22:26 +0200)]
Use override for ARCH in make.inc

in case a conflicting setting of ARCH (for architecture) gets pulled in from the environment
(originally suggested by dloghin in #1753)

5 years agoMerge pull request #1798 from martin-frbg/cmake-avx512
Martin Kroeker [Mon, 8 Oct 2018 19:15:17 +0000 (21:15 +0200)]
Merge pull request #1798 from martin-frbg/cmake-avx512

Add -march=skylake-avx512 when required

5 years agoAdd -march=skylake-avx512 when required
Martin Kroeker [Mon, 8 Oct 2018 17:18:12 +0000 (19:18 +0200)]
Add -march=skylake-avx512 when required

fixes #1797

5 years agodgemm/skylakex: replace discrete mul/add with fma
Arjan van de Ven [Sat, 6 Oct 2018 23:13:26 +0000 (23:13 +0000)]
dgemm/skylakex: replace discrete mul/add with fma

very minor gains since it's not super hot code, but general principles

5 years agoAdd vector optimizations for ncopy as well for dgemm/skylakex
Arjan van de Ven [Sat, 6 Oct 2018 21:18:12 +0000 (21:18 +0000)]
Add vector optimizations for ncopy as well for dgemm/skylakex

5 years agoadd a skylakex optimized dgemm beta function
Arjan van de Ven [Sat, 6 Oct 2018 16:36:26 +0000 (16:36 +0000)]
add a skylakex optimized dgemm beta function

5 years agoMerge pull request #1791 from dev-zero/develop
Martin Kroeker [Sat, 6 Oct 2018 14:29:29 +0000 (16:29 +0200)]
Merge pull request #1791 from dev-zero/develop

fix parallel build issues with APFS/HFS+/ext2/3 in netlib-lapack

5 years agodgemm/avx512 simplify and speed up the 4x4 kernel
Arjan van de Ven [Sat, 6 Oct 2018 14:12:32 +0000 (14:12 +0000)]
dgemm/avx512 simplify and speed up the 4x4 kernel

5 years agoundo slow dgemm/skylake microoptimization
Arjan van de Ven [Sat, 6 Oct 2018 14:00:37 +0000 (14:00 +0000)]
undo slow dgemm/skylake microoptimization

the compare is more costly than the work

5 years agoAdd optimized *copy versions for skylakex
Arjan van de Ven [Sat, 6 Oct 2018 13:47:20 +0000 (13:47 +0000)]
Add optimized *copy versions for skylakex

Add optimized n/t copy versions for skylakex; in the patch the
tcopy is also rewritten using intrinsics; the ncopy file
will be worked on in a future commit

5 years agoMerge pull request #6 from xianyi/develop
Martin Kroeker [Sat, 6 Oct 2018 12:36:36 +0000 (14:36 +0200)]
Merge pull request #6 from xianyi/develop

merge develop

5 years agoAdd SYMBOLPREFIX and -SUFFIX options and improve help output
Martin Kroeker [Sat, 6 Oct 2018 12:28:04 +0000 (14:28 +0200)]
Add SYMBOLPREFIX and -SUFFIX options and improve help output

5 years agofix parallel build issues with APFS/HFS+/ext2/3 in netlib-lapack
Tiziano Müller [Sat, 6 Oct 2018 12:10:02 +0000 (14:10 +0200)]
fix parallel build issues with APFS/HFS+/ext2/3 in netlib-lapack

The problem is that OpenBLAS sets the LAPACKE_LIB and the TMGLIB to the
same object and uses the `ar` feature to update the archive file. If the
underlying filesystem does not have sub-second timestamp resolution and
the system is fast enough (or `ccache` is used), the timestamp of the
builds which should be added to the previously generated archive is the
same as the archive file itself and therefore `make` does not update the
archive.

Since OpenBLAS takes care to not run the different targets updating the
archive in parallel, the easiest solution is to declare the respective
targets `.PHONY`, forcing `make` to always update them.

fixes #1682

5 years agoMerge pull request #1789 from brada4/develop
Martin Kroeker [Fri, 5 Oct 2018 18:42:37 +0000 (20:42 +0200)]
Merge pull request #1789 from brada4/develop

update travis alpine chroot with avx512 intrinsics headers

5 years agoMerge pull request #1788 from fenrus75/avx512-8x16
Martin Kroeker [Fri, 5 Oct 2018 18:40:38 +0000 (20:40 +0200)]
Merge pull request #1788 from fenrus75/avx512-8x16

skylake dgemm: Add a 16x8 kernel

5 years agoAdd a 24x8 kernel to the skylakex dgemm implementation
Arjan van de Ven [Fri, 5 Oct 2018 13:22:21 +0000 (13:22 +0000)]
Add a 24x8 kernel to the skylakex dgemm implementation

Minor gains for small matrixes, but at 512x512 and above the gain
gets more significant.

5 years agoskylake dgemm: Add a 16x8 kernel
Arjan van de Ven [Fri, 5 Oct 2018 11:49:43 +0000 (11:49 +0000)]
skylake dgemm: Add a 16x8 kernel

The next step for the avx512 dgemm code is adding a 16x8 kernel.
In the 8x8 kernel, each FMA has a matching load (the broadcast);
in the 16x8 kernel we can reuse this load for 2 FMAs, which
in turn reduces pressure on the load ports of the CPU and gives
a nice performance boost (in the 25% range).

5 years agoupdate travis alpine chroot with avx512 intrinsics headers
Andrew [Fri, 5 Oct 2018 12:47:55 +0000 (15:47 +0300)]
update travis alpine chroot with avx512 intrinsics headers

5 years agoupdate travis alpine chroot with avx512 intrinsics headers
Andrew [Fri, 5 Oct 2018 12:41:52 +0000 (15:41 +0300)]
update travis alpine chroot with avx512 intrinsics headers

5 years agoMerge pull request #1785 from brada4/develop
Martin Kroeker [Fri, 5 Oct 2018 06:25:38 +0000 (08:25 +0200)]
Merge pull request #1785 from brada4/develop

address #1782 2nd loop

5 years agoMerge pull request #1784 from fenrus75/dgemm-avx512
Martin Kroeker [Fri, 5 Oct 2018 06:03:27 +0000 (08:03 +0200)]
Merge pull request #1784 from fenrus75/dgemm-avx512

Create a AVX512 enabled version of DGEMM

5 years agoFunction name needs to be CNAME, set from outside to allow suffixing for dynamic_arch
Martin Kroeker [Thu, 4 Oct 2018 17:14:59 +0000 (19:14 +0200)]
Function name needs to be CNAME, set from outside to allow suffixing for dynamic_arch

5 years agoMerge pull request #1787 from jeromerobert/develop
Martin Kroeker [Thu, 4 Oct 2018 16:41:47 +0000 (18:41 +0200)]
Merge pull request #1787 from jeromerobert/develop

Fix unknown type name __WAIT_STATUS on RHEL5

5 years agoFix unknown type name __WAIT_STATUS on RHEL5
Jerome Robert [Thu, 4 Oct 2018 10:27:44 +0000 (12:27 +0200)]
Fix unknown type name __WAIT_STATUS on RHEL5

With glibc 2.5 one must have #define _XOPEN_SOURCE >= 500 to use wait.
But reading glibc code this is actually needed only if stdlib.h was
included before sys/wait.h. This was the case here through
openblas_utest.h. So changing include fix compilation on RHEL5 and
should ne hurt with more recent distro.

* Problem found when using with gcc 5.5 and 4.7.2 on RHEL5/CENTOS5
* Fix #1519

5 years agoMerge pull request #1786 from martin-frbg/immintrin
Martin Kroeker [Thu, 4 Oct 2018 07:07:09 +0000 (09:07 +0200)]
Merge pull request #1786 from martin-frbg/immintrin

Check for Immintrin.h presence in the AVX512 compatibility test as well

5 years agoCheck availability of immintrin.h in the AVX512 compatibility test
Martin Kroeker [Thu, 4 Oct 2018 05:36:49 +0000 (07:36 +0200)]
Check availability of immintrin.h in the AVX512 compatibility test

5 years agoCheck availability of immintrin.h in the AVX512 compatibility test
Martin Kroeker [Thu, 4 Oct 2018 05:35:30 +0000 (07:35 +0200)]
Check availability of immintrin.h in the AVX512 compatibility test

5 years agoaddress #1782 2nd loop
Andrew [Wed, 3 Oct 2018 19:20:50 +0000 (21:20 +0200)]
address #1782 2nd loop

5 years agoCreate a AVX512 enabled version of DGEMM
Arjan van de Ven [Wed, 3 Oct 2018 14:45:25 +0000 (14:45 +0000)]
Create a AVX512 enabled version of DGEMM

This patch adds dgemm_kernel_4x8_skylakex.c which is
* dgemm_kernel_4x8_haswell.s converted to C + intrinsics
* 8x8 support added
* 8x8 kernel implemented using AVX512

Performance is a work in progress, but already shows a 10% - 20%
increase for a wide range of matrix sizes.

5 years agoMerge pull request #1780 from martin-frbg/issue1774-2
Martin Kroeker [Sat, 29 Sep 2018 07:27:47 +0000 (09:27 +0200)]
Merge pull request #1780 from martin-frbg/issue1774-2

Convert fldmia/fstmia instructions to UAL syntax for clang7

5 years agoConvert fldmia/fstmia instructions to UAL syntax for clang7
Martin Kroeker [Fri, 28 Sep 2018 21:05:15 +0000 (23:05 +0200)]
Convert fldmia/fstmia instructions to UAL syntax for clang7

second part of fix for #1774, containing files missed in #1775

5 years agoMerge pull request #1778 from fengrl/develop
Martin Kroeker [Wed, 26 Sep 2018 09:14:58 +0000 (11:14 +0200)]
Merge pull request #1778 from fengrl/develop

test_axpy work error on LOONGSON3A platform #1777

5 years agotest_axpy work error on LOONGSON3A platform #1777
fengruilin [Wed, 26 Sep 2018 07:14:04 +0000 (15:14 +0800)]
test_axpy work error on LOONGSON3A platform #1777

5 years agoMerge pull request #1775 from martin-frbg/issue1774
Martin Kroeker [Tue, 25 Sep 2018 16:58:39 +0000 (18:58 +0200)]
Merge pull request #1775 from martin-frbg/issue1774

Convert fldmia/fstmia instructions to UAL syntax for clang7

5 years agoConvert fldmia/fstmia instructions to UAL syntax for clang7
Martin Kroeker [Tue, 25 Sep 2018 07:41:58 +0000 (09:41 +0200)]
Convert fldmia/fstmia instructions to UAL syntax for clang7

fixes #1774

5 years agoMerge pull request #1773 from martin-frbg/issue1767
Martin Kroeker [Sun, 23 Sep 2018 21:25:15 +0000 (23:25 +0200)]
Merge pull request #1773 from martin-frbg/issue1767

Include thread numbers in failure message from blas_thread_init

5 years agoInclude thread numbers in failure message from blas_thread_init
Martin Kroeker [Sat, 22 Sep 2018 12:00:15 +0000 (14:00 +0200)]
Include thread numbers in failure message from blas_thread_init

to aid in debugging cases like #1767

5 years agoMerge pull request #1771 from staticfloat/sf/ldflags
Martin Kroeker [Sat, 22 Sep 2018 11:11:39 +0000 (13:11 +0200)]
Merge pull request #1771 from staticfloat/sf/ldflags

Add `$(LDFLAGS)` to `$(CC)` and `$(FC)` invocations within `exports/Makefile`

5 years agoDocument the stub status of the QUAD_PRECiSION code (#1772)
Martin Kroeker [Sat, 22 Sep 2018 10:31:37 +0000 (12:31 +0200)]
Document the stub status of the QUAD_PRECiSION code (#1772)

* Document the stub status of the QUAD_PRECiSION code inherited from GotoBLAS2

in response to #1769

5 years agoAdd `$(LDFLAGS)` to `$(CC)` and `$(FC)` invocations within `exports/Makefile`
Elliot Saba [Fri, 21 Sep 2018 09:19:51 +0000 (09:19 +0000)]
Add `$(LDFLAGS)` to `$(CC)` and `$(FC)` invocations within `exports/Makefile`

5 years agoMerge pull request #1765 from martin-frbg/issue1761
Martin Kroeker [Wed, 19 Sep 2018 20:02:21 +0000 (22:02 +0200)]
Merge pull request #1765 from martin-frbg/issue1761

Do not use the new TLS-enabled memory allocator for non-threaded builds, and disable TLS by default in gmake as well

5 years agoMerge pull request #1764 from yurivict/64-suffix
Martin Kroeker [Wed, 19 Sep 2018 16:16:38 +0000 (18:16 +0200)]
Merge pull request #1764 from yurivict/64-suffix

Allow to install the 'interface64' version concurrently with the regular version

5 years agoMerge pull request #1762 from martin-frbg/issue1710-2
Martin Kroeker [Wed, 19 Sep 2018 16:16:21 +0000 (18:16 +0200)]
Merge pull request #1762 from martin-frbg/issue1710-2

Add explicit casts to silence compiler warnings

5 years agoFix default settings - USE_TLS and USE_SIMPLE_THREADED_LEVEL3 should both be off
Martin Kroeker [Wed, 19 Sep 2018 16:08:31 +0000 (18:08 +0200)]
Fix default settings - USE_TLS and USE_SIMPLE_THREADED_LEVEL3 should both be off

5 years agoCatch inadvertent USE_TLS=0 declaration
Martin Kroeker [Wed, 19 Sep 2018 16:03:43 +0000 (18:03 +0200)]
Catch inadvertent USE_TLS=0 declaration

for #1766

5 years agoDo not use the new TLS code for non-threaded builds even if USE_TLS is set
Martin Kroeker [Sun, 16 Sep 2018 10:43:36 +0000 (12:43 +0200)]
Do not use the new TLS code for non-threaded builds even if USE_TLS is set

Workaround for #1761 as that exposed a problem in the new code (which was intended to speed up multithreaded code only anyway).