Zhang Xianyi [Mon, 10 Aug 2015 19:10:44 +0000 (14:10 -0500)]
Use cmake to build OpenBLAS GENERIC Target on MSVC x86 64-bit.
Disable CBLAS and LAPACK.
Zhang Xianyi [Tue, 4 Aug 2015 04:59:01 +0000 (23:59 -0500)]
Merge branch 'develop' into cmake
Zhang Xianyi [Tue, 4 Aug 2015 04:55:56 +0000 (23:55 -0500)]
Use pure C generic target on x86 and x86_64.
make TARGET=GENERIC
?gemm3m is unimplemented on generic target.
Zhang Xianyi [Mon, 3 Aug 2015 20:37:05 +0000 (04:37 +0800)]
Fixed cmake building bugs on Linux. Disable LAPACK by default.
Zhang Xianyi [Tue, 21 Jul 2015 20:07:27 +0000 (04:07 +0800)]
Merge branch 'hpanderson_cmake' into cmake
Zhang Xianyi [Tue, 21 Jul 2015 20:06:39 +0000 (04:06 +0800)]
Merge branch 'cmake' of https://github.com/hpanderson/OpenBLAS into hpanderson_cmake
Zhang Xianyi [Fri, 26 Jun 2015 03:54:41 +0000 (11:54 +0800)]
Fix blas lock bug on AArch64.
Zhang Xianyi [Tue, 23 Jun 2015 02:54:51 +0000 (21:54 -0500)]
Merge pull request #595 from tanderson92/fixTests
Fix test execution when USE_OPENMP=0
wernsaar [Sat, 13 Jun 2015 14:44:48 +0000 (16:44 +0200)]
Merge pull request #596 from wernsaar/develop
optimizations for haswell
Werner Saar [Sat, 13 Jun 2015 14:16:29 +0000 (16:16 +0200)]
added optimized dtrmm_kernel for haswell
Werner Saar [Sat, 13 Jun 2015 08:28:27 +0000 (10:28 +0200)]
modified haswell parameter dgemm_unroll_n
Thomas Anderson [Sat, 13 Jun 2015 06:52:07 +0000 (23:52 -0700)]
Fix test execution when USE_OPENMP=0
The standard way to disable OpenMP support is to set USE_OPENMP=0,
as indicated by other checks to see if USE_OPENMP equals 1. The
problem is obviously then that `ifdef USE_OPENMP` is very much not
what we want to test for. This causes tests to fail when no OpenMP
library is installed.
Zhang Xianyi [Mon, 8 Jun 2015 15:53:50 +0000 (10:53 -0500)]
Fix #593. Change MACOSX_DEPLOYMENT_TARGET to 10.6.
wernsaar [Mon, 8 Jun 2015 12:22:02 +0000 (14:22 +0200)]
Merge pull request #592 from wernsaar/develop
added benchmark scripts
Werner Saar [Mon, 8 Jun 2015 12:06:38 +0000 (14:06 +0200)]
added benchmark scripts for numpy, octave and R
Werner Saar [Mon, 8 Jun 2015 10:58:38 +0000 (12:58 +0200)]
updated geev benchmark
wernsaar [Wed, 3 Jun 2015 10:14:09 +0000 (12:14 +0200)]
Merge pull request #589 from wernsaar/develop
small modification of gemm.c
Werner Saar [Wed, 3 Jun 2015 07:11:51 +0000 (09:11 +0200)]
small modification of gemm.c
wernsaar [Tue, 2 Jun 2015 13:29:49 +0000 (15:29 +0200)]
Merge pull request #587 from wernsaar/develop
added gesv benchmark
Werner Saar [Tue, 2 Jun 2015 11:35:49 +0000 (13:35 +0200)]
added gesv benchmark
wernsaar [Sun, 31 May 2015 13:01:54 +0000 (15:01 +0200)]
Merge pull request #585 from wernsaar/develop
bugfix for benchmark Makefile on MAC
Werner Saar [Sun, 31 May 2015 12:16:51 +0000 (14:16 +0200)]
bugfix for Makefile on mac
wernsaar [Fri, 29 May 2015 11:27:20 +0000 (13:27 +0200)]
Merge pull request #584 from wernsaar/develop
bugfixes, to build benchmarks with mingw on Windows OS
Werner Saar [Fri, 29 May 2015 10:56:22 +0000 (12:56 +0200)]
bugfixes, to build benchmarks with mingw on Windows OS
wernsaar [Sat, 23 May 2015 10:58:15 +0000 (12:58 +0200)]
Merge pull request #581 from wernsaar/develop
bugfix for arm locking
Werner Saar [Sat, 23 May 2015 09:40:40 +0000 (11:40 +0200)]
bugfix for arm locking
Werner Saar [Sat, 23 May 2015 08:58:38 +0000 (10:58 +0200)]
smp lock bugfix
wernsaar [Sat, 23 May 2015 07:46:39 +0000 (09:46 +0200)]
Merge pull request #580 from wernsaar/develop
added blas level1 swap benchmark
Werner Saar [Thu, 21 May 2015 06:51:42 +0000 (08:51 +0200)]
added blas level1 swap benchmark
Zhang Xianyi [Thu, 21 May 2015 02:57:27 +0000 (21:57 -0500)]
Support Android NDK armeabi-v7a-hard ABI. (-mfloat-abi=hard)
e.g.
make HOSTCC=gcc CC=arm-linux-androideabi-gcc NO_LAPACK=1 TARGET=ARMV7
In Android NDK, it uses armeabi-v7a-hard ABI.
TARGET_CFLAGS += -mhard-float -D_NDK_MATH_NO_SOFTFP=1
TARGET_LDFLAGS += -Wl,--no-warn-mismatch -lm_hard
For more information, please check hard-float example at
android_ndk/tests/device/hard-float/jni/.
wernsaar [Wed, 20 May 2015 09:56:02 +0000 (11:56 +0200)]
Merge pull request #578 from wernsaar/develop
added blas level1 copy benchmark
Werner Saar [Wed, 20 May 2015 09:05:00 +0000 (11:05 +0200)]
added blas level1 copy benchmark
Zhang Xianyi [Tue, 19 May 2015 17:04:45 +0000 (12:04 -0500)]
Fix f_check bug.
wernsaar [Tue, 19 May 2015 08:59:24 +0000 (10:59 +0200)]
Merge pull request #577 from wernsaar/develop
Bugfix for armv6 memory barrier
Werner Saar [Tue, 19 May 2015 08:43:12 +0000 (10:43 +0200)]
Ref #574: Bugfix for armv6 memory barrier
Zhang Xianyi [Tue, 19 May 2015 05:01:04 +0000 (00:01 -0500)]
1) Refs #575. Remove g77 from compiler list.
2) If OpenBLAS cannot find Fortran compiler, it will only build BLAS
(without LAPACK).
wernsaar [Mon, 18 May 2015 11:47:38 +0000 (13:47 +0200)]
Merge pull request #572 from wernsaar/develop
added optimized cscal and zscal functions for steamroller
Werner Saar [Mon, 18 May 2015 10:40:07 +0000 (12:40 +0200)]
added optimized cscal and zscal kernels for steamroller
Werner Saar [Mon, 18 May 2015 08:50:57 +0000 (10:50 +0200)]
added optimized cscal and zscal kernels for steamroller and piledriver
Werner Saar [Mon, 18 May 2015 06:46:06 +0000 (08:46 +0200)]
added optimized cscal kernel for sandybridge
Werner Saar [Mon, 18 May 2015 05:33:52 +0000 (07:33 +0200)]
added optimized cscal kernel for bulldozer
wernsaar [Sun, 17 May 2015 12:09:14 +0000 (14:09 +0200)]
Merge pull request #571 from wernsaar/develop
added optimized cscal and zscal functions
Werner Saar [Sun, 17 May 2015 11:44:09 +0000 (13:44 +0200)]
added optimized cscal kernel for haswell
Werner Saar [Sun, 17 May 2015 09:45:19 +0000 (11:45 +0200)]
added optimized zscal kernel for bulldozer
Werner Saar [Sat, 16 May 2015 14:41:45 +0000 (16:41 +0200)]
added optimized zscal kernel for haswell
Zhang Xianyi [Wed, 13 May 2015 21:16:30 +0000 (16:16 -0500)]
Add AMD Excavator target.
wernsaar [Wed, 13 May 2015 11:48:08 +0000 (13:48 +0200)]
Merge pull request #568 from wernsaar/develop
added optimized dscal kernel
Werner Saar [Wed, 13 May 2015 11:31:26 +0000 (13:31 +0200)]
bugfix: added static to functions
Werner Saar [Wed, 13 May 2015 11:05:35 +0000 (13:05 +0200)]
added optimized dscal kernel for piledriver
Werner Saar [Wed, 13 May 2015 10:14:39 +0000 (12:14 +0200)]
optimized dscal kernel for increment != 1
Werner Saar [Tue, 12 May 2015 15:19:58 +0000 (17:19 +0200)]
added optimized dscal kernel for haswell
Werner Saar [Tue, 12 May 2015 14:27:43 +0000 (16:27 +0200)]
added optimized dscal kernel for sandybridge
Werner Saar [Tue, 12 May 2015 10:28:44 +0000 (12:28 +0200)]
added optimized dscal kernel for bulldozer
Zhang Xianyi [Tue, 12 May 2015 01:59:12 +0000 (20:59 -0500)]
Merge pull request #566 from powderluv/develop
Fix build with ALLOC_SHM=0 (Android NDK)
powderluv [Sun, 10 May 2015 07:10:26 +0000 (00:10 -0700)]
Fix build with ALLOC_SHM=0 (Android NDK)
Refactor such that you can build with ALLOC_SHM=0. HughTLB
implicity depends on ALLOC_SHM=1. This patch allows
building for Android NDK r10d.
Zhang Xianyi [Thu, 7 May 2015 21:33:17 +0000 (05:33 +0800)]
Refs #532. Improve gemv paralel with small m and large n case.
Splite the matrix and reduction.
Zhang Xianyi [Thu, 7 May 2015 05:06:53 +0000 (13:06 +0800)]
Refs #565. Fix the bug of generate FEXTRALIB.
Zhang Xianyi [Thu, 7 May 2015 04:52:14 +0000 (12:52 +0800)]
Refs #565. Merge branch 'andreasnoack-anj/bench' into develop
Andreas Noack [Thu, 7 May 2015 01:52:34 +0000 (21:52 -0400)]
Add vecLib benchmarks
wernsaar [Wed, 6 May 2015 09:10:31 +0000 (11:10 +0200)]
Merge pull request #564 from wernsaar/develop
Use only 1 thread in trsm if m or n < 2*GEMM_MULTITHREAD_THRESHOLD
Werner Saar [Wed, 6 May 2015 08:41:53 +0000 (10:41 +0200)]
use only 1 thread if m or n < 2*GEMM_MULTITHREAD_THRESHOLD
Werner Saar [Wed, 6 May 2015 07:21:19 +0000 (09:21 +0200)]
added loops to trsm.c
wernsaar [Tue, 5 May 2015 10:13:35 +0000 (12:13 +0200)]
Merge pull request #563 from wernsaar/develop
Bugfix for gemm3m tests
Werner Saar [Tue, 5 May 2015 09:58:59 +0000 (11:58 +0200)]
bugfix for gemm3m tests
Werner Saar [Tue, 5 May 2015 09:39:43 +0000 (11:39 +0200)]
removed gemm3m functions from normal checks
wernsaar [Mon, 4 May 2015 09:11:13 +0000 (11:11 +0200)]
Merge pull request #561 from wernsaar/develop
updated dgemv_n sgemv_n kernels
Werner Saar [Thu, 30 Apr 2015 12:38:06 +0000 (14:38 +0200)]
updated dgemv_n kernel for nehalem
Werner Saar [Thu, 30 Apr 2015 10:11:39 +0000 (12:11 +0200)]
optimized dgemv_n kernel for haswell
Zhang Xianyi [Wed, 29 Apr 2015 16:36:47 +0000 (11:36 -0500)]
Merge pull request #560 from sebastien-villemot/develop
Fix detection of ARM architectures in c_check.
Sébastien Villemot [Wed, 29 Apr 2015 16:14:21 +0000 (18:14 +0200)]
Fix detection of ARM architectures in c_check.
This is necessary to avoid the false detection of a cross-compiling environment.
wernsaar [Tue, 28 Apr 2015 15:30:16 +0000 (17:30 +0200)]
Merge pull request #558 from wernsaar/develop
optimizations for sandybridge
Werner Saar [Tue, 28 Apr 2015 14:58:11 +0000 (16:58 +0200)]
optimized dger kernel for sandybridge
Werner Saar [Tue, 28 Apr 2015 13:33:38 +0000 (15:33 +0200)]
added optimized sger kernel for sandybridge
Werner Saar [Tue, 28 Apr 2015 08:18:32 +0000 (10:18 +0200)]
optimized saxpy and daxpy for sandybridge
Zhang Xianyi [Sat, 25 Apr 2015 13:11:36 +0000 (08:11 -0500)]
Merge pull request #554 from wernsaar/develop
added benchmarks for zgeru and cgeru
Werner Saar [Sat, 25 Apr 2015 12:53:07 +0000 (14:53 +0200)]
add benchmarks for zgeru and cgeru
Zhang Xianyi [Fri, 24 Apr 2015 19:12:12 +0000 (14:12 -0500)]
Merge pull request #552 from jeromerobert/develop
gemv: Ensure stack buffer is large enough to handle memory alignment
Werner Saar [Fri, 24 Apr 2015 12:30:44 +0000 (14:30 +0200)]
bugfixes: replaced int with BLASLONG
wernsaar [Fri, 24 Apr 2015 11:57:48 +0000 (13:57 +0200)]
Merge pull request #553 from wernsaar/develop
optimized some blas level1 kernels for increments != 1
Werner Saar [Fri, 24 Apr 2015 11:13:20 +0000 (13:13 +0200)]
optimized sdot.c for increments != 1
Werner Saar [Fri, 24 Apr 2015 09:52:59 +0000 (11:52 +0200)]
optimized saxpy.c for increments != 1
Werner Saar [Fri, 24 Apr 2015 09:39:17 +0000 (11:39 +0200)]
optimized daxpy kernel for increments != 1
Werner Saar [Fri, 24 Apr 2015 08:56:55 +0000 (10:56 +0200)]
optimized ddot.c for increments != 1
Jerome Robert [Tue, 21 Apr 2015 08:12:01 +0000 (10:12 +0200)]
gemv: Ensure stack buffer is large enough to handle memory alignment
Ref #478
wernsaar [Thu, 23 Apr 2015 11:27:38 +0000 (13:27 +0200)]
Merge pull request #550 from wernsaar/develop
added optimized ssymv kernels for haswell and sandybridge
Werner Saar [Thu, 23 Apr 2015 10:19:24 +0000 (12:19 +0200)]
added optimized ssymv kernels for sandybridge
Werner Saar [Thu, 23 Apr 2015 08:23:13 +0000 (10:23 +0200)]
added optimized ssymv kernels for haswell
wernsaar [Wed, 22 Apr 2015 10:36:13 +0000 (12:36 +0200)]
Merge pull request #549 from wernsaar/develop
added optimized dsymv kernels for haswell and sandybridge
Werner Saar [Wed, 22 Apr 2015 10:09:43 +0000 (12:09 +0200)]
added optimized dsymv kernels for sandybridge
Werner Saar [Wed, 22 Apr 2015 08:42:50 +0000 (10:42 +0200)]
added optimized dsymv kernels for haswell
Zhang Xianyi [Tue, 21 Apr 2015 04:22:40 +0000 (23:22 -0500)]
Refs #478,#482, Enable stack alloc for s/dgemv_t.(revert 9798491)
Werner Saar [Sun, 19 Apr 2015 09:24:07 +0000 (11:24 +0200)]
added asum benchmark
Werner Saar [Sat, 18 Apr 2015 06:41:41 +0000 (08:41 +0200)]
added scal benchmark
wernsaar [Thu, 16 Apr 2015 09:36:51 +0000 (11:36 +0200)]
Merge pull request #546 from wernsaar/develop
added optimized zaxpy-kernels
Werner Saar [Thu, 16 Apr 2015 09:19:37 +0000 (11:19 +0200)]
added optimized zaxpy-kernels
Zhang Xianyi [Wed, 15 Apr 2015 16:18:14 +0000 (11:18 -0500)]
Merge pull request #543 from jeromerobert/develop
Fix a buffer overflow with MAX_STACK_ALLOC size in dgemv_t
wernsaar [Wed, 15 Apr 2015 15:04:02 +0000 (17:04 +0200)]
Merge pull request #544 from wernsaar/develop
Optimized caxpy-kernels
Werner Saar [Wed, 15 Apr 2015 14:29:25 +0000 (16:29 +0200)]
added optimized caxpy-kernel for sandybridge
Werner Saar [Wed, 15 Apr 2015 13:16:31 +0000 (15:16 +0200)]
added optimized caxpy-kernel for haswell
Werner Saar [Wed, 15 Apr 2015 11:49:23 +0000 (13:49 +0200)]
added optimized caxpy-kernel for steamroller