Martin Kroeker [Fri, 9 Feb 2018 22:06:50 +0000 (23:06 +0100)]
Remove debug printfs
Martin Kroeker [Fri, 9 Feb 2018 22:00:03 +0000 (23:00 +0100)]
Keep the flag handling separate from the scaling loops
Fixes #1452 and is more in line with how ATLAS does it. The earlier fix from #356 only moved the bug elsewhere, but we will never want the iterative rescaling to change the dflag setting and variable associations with each cycle.
Martin Kroeker [Thu, 1 Feb 2018 07:28:53 +0000 (08:28 +0100)]
Merge pull request #1445 from quickwritereader/develop
small fix inside ifdef z13mvc . (z13mvc code is not used in production)
Abdelrauf [Thu, 1 Feb 2018 00:17:04 +0000 (16:17 -0800)]
Merge branch 'develop' into develop
Abdelrauf [Wed, 31 Jan 2018 15:49:38 +0000 (07:49 -0800)]
small fix inside ifdef z13mvc . (z13mvc code is not used in production)
Martin Kroeker [Wed, 31 Jan 2018 22:49:01 +0000 (23:49 +0100)]
Merge pull request #1443 from martin-frbg/sparcfix
Also #define SPARC in config.h when autodetecting
Martin Kroeker [Wed, 31 Jan 2018 22:48:47 +0000 (23:48 +0100)]
Merge pull request #1440 from quickwritereader/develop
small corrections
Martin Kroeker [Wed, 31 Jan 2018 22:48:34 +0000 (23:48 +0100)]
Merge pull request #1419 from brada4/develop
Initialize unitialized values for repeated calls
Martin Kroeker [Wed, 31 Jan 2018 21:02:00 +0000 (22:02 +0100)]
Also #define SPARC in config.h when autodetecting
Fixes #1442
Abdelrauf [Wed, 31 Jan 2018 15:49:38 +0000 (07:49 -0800)]
small fix
small fix inside ifdef z13mvc . (z13mvc code is not used in production)
Martin Kroeker [Sat, 27 Jan 2018 13:26:49 +0000 (14:26 +0100)]
Merge pull request #1434 from xoviat/flang-wall
CMake: Remove unused wall option when FC=flang
Martin Kroeker [Sat, 27 Jan 2018 09:04:39 +0000 (10:04 +0100)]
Merge pull request #1436 from martin-frbg/cmaketrmm
Make USE_TRMM depend on TARGET_CORE not TARGET
the mslm [Sat, 27 Jan 2018 02:20:36 +0000 (18:20 -0800)]
zscal (case: real alpha=0 ) mikrokernel shift&mem fix , da_i as input reg. small typo fixes
Martin Kroeker [Fri, 26 Jan 2018 22:20:00 +0000 (23:20 +0100)]
Make USE_TRMM depend on TARGET_CORE not TARGET
Fixes #1432 (and possibly other DTRMM-related failures on Haswell and related architectures when built with cmake)
xoviat [Fri, 26 Jan 2018 20:09:48 +0000 (14:09 -0600)]
CMake: Remove unused wall option when FC=flang
Martin Kroeker [Wed, 24 Jan 2018 08:02:55 +0000 (09:02 +0100)]
Merge pull request #1429 from martin-frbg/override_omp
When forcing USE_THREAD to zero, override USE_OPENMP as well
Martin Kroeker [Tue, 23 Jan 2018 20:33:21 +0000 (21:33 +0100)]
When forcing USE_THREAD=0, override USE_OPENMP as well
This avoids an error exit a few lines down as USE_THREAD=0 conflicts with USE_OPENMP=1
Martin Kroeker [Sat, 20 Jan 2018 17:20:42 +0000 (18:20 +0100)]
Merge pull request #1417 from xoviat/openblas-library-name
CMake: Use the correct library name on windows
Martin Kroeker [Sat, 20 Jan 2018 16:34:54 +0000 (17:34 +0100)]
Merge pull request #1426 from quickwritereader/develop
(Z13 ) Blas1 mikrokernels can be inlined by gcc. Refactoring,fixes,tunings
Andrew [Sat, 20 Jan 2018 10:42:31 +0000 (11:42 +0100)]
take out unused variables
the mslm [Thu, 18 Jan 2018 02:05:44 +0000 (18:05 -0800)]
Blas1 mikrokernels can be inlined by gcc. Refactoring ( symbolic operan
names). Some fixes and tunings
Andrew [Fri, 19 Jan 2018 22:17:43 +0000 (23:17 +0100)]
core.IdenticalExpr clang501 checker
Andrew [Fri, 19 Jan 2018 22:15:58 +0000 (23:15 +0100)]
core.IdenticalExpr clang501 checker
Andrew [Fri, 19 Jan 2018 22:11:12 +0000 (23:11 +0100)]
add missing brackets to silence indentation warnings gcc721
Andrew [Fri, 12 Jan 2018 21:35:48 +0000 (22:35 +0100)]
add missing bracket for old glibc (cppcheck)
Andrew [Fri, 12 Jan 2018 21:35:00 +0000 (22:35 +0100)]
Initialize values to silence cppcheck
Andrew [Fri, 12 Jan 2018 21:33:41 +0000 (22:33 +0100)]
Initialize uninitialized variables (cppcheck)
xoviat [Thu, 11 Jan 2018 17:34:53 +0000 (11:34 -0600)]
CMake: Use the correct library name on windows
FindBLAS and FindLAPACK don't correctly detect the OpenBLAS library
name on windows. I am unsure why this was originally set this way, but
it has caused me some trouble.
Martin Kroeker [Thu, 11 Jan 2018 07:35:02 +0000 (08:35 +0100)]
Merge pull request #1415 from quickwritereader/develop
(Z systems Z13) small fixes, some (i(dz)amin,i(dz)amax,(dz)dot,(dz)asum) mikrokernels…
Abdelrauf [Fri, 13 Oct 2017 15:29:27 +0000 (19:29 +0400)]
small fixes, some (i(dz)amin,i(dz)amax,(dz)dot,(dz)asum) mikrokernels can be inlined
Martin Kroeker [Sat, 6 Jan 2018 19:02:46 +0000 (20:02 +0100)]
Merge pull request #1410 from brada4/develop
Address warnings #1357
Martin Kroeker [Sat, 6 Jan 2018 19:01:35 +0000 (20:01 +0100)]
Merge pull request #1411 from christoph-conrads/fix-pkgconfig-path-escapes
Make: escape paths to pkg-config file
Christoph Conrads [Fri, 5 Jan 2018 22:08:55 +0000 (17:08 -0500)]
Make: escape paths to pkg-config file
Add double quotes around the path to the pkg-config file so that a path
containing whitespace does not break the build.
Andrew [Tue, 2 Jan 2018 13:38:50 +0000 (14:38 +0100)]
LAPACK helpers in C that need care too
Andrew [Mon, 1 Jan 2018 19:57:12 +0000 (20:57 +0100)]
address last warnings as seen by gcc7
Andrew [Mon, 1 Jan 2018 19:56:26 +0000 (20:56 +0100)]
remove surplus parentheses to silence clang5
Andrew [Mon, 1 Jan 2018 19:54:39 +0000 (20:54 +0100)]
Eliminate remaining unused results in kernels (clang5 analyzer)
Martin Kroeker [Sun, 31 Dec 2017 19:18:48 +0000 (20:18 +0100)]
Merge pull request #1409 from martin-frbg/issue1292-2
Tag %1 and %2 as both input and output operands
Martin Kroeker [Sun, 31 Dec 2017 17:03:36 +0000 (18:03 +0100)]
Tag %1 and %2 as both input and output operands
fix from #1292 extended to the other gemv microkernels
Martin Kroeker [Sat, 30 Dec 2017 13:52:21 +0000 (14:52 +0100)]
Merge pull request #1408 from xoviat/flang-ninja
Appveyor: speed up fortran builds
Martin Kroeker [Sat, 30 Dec 2017 13:52:03 +0000 (14:52 +0100)]
Merge pull request #1406 from martin-frbg/issue1292
Tag %1 and %2 as both input and output
Martin Kroeker [Sat, 30 Dec 2017 13:51:34 +0000 (14:51 +0100)]
Merge pull request #1403 from brada4/develop
Address few more warnings
xoviat [Sat, 30 Dec 2017 01:58:35 +0000 (19:58 -0600)]
Appveyor: enable building fortran with ninja
Martin Kroeker [Fri, 29 Dec 2017 22:56:41 +0000 (23:56 +0100)]
Tag %1 and %2 as both input and output
The inline assembly modifies its input operands, so mark them as output to avoid surprises with optimization. Fixes #1292
Andrew [Tue, 26 Dec 2017 08:24:24 +0000 (09:24 +0100)]
initialize potentially unitialized variables (clang5)
Andrew [Thu, 21 Dec 2017 23:56:35 +0000 (00:56 +0100)]
fix couple of dead assignment warnings
Andrew [Thu, 21 Dec 2017 23:55:40 +0000 (00:55 +0100)]
remove unused buffer
Martin Kroeker [Thu, 21 Dec 2017 22:36:52 +0000 (23:36 +0100)]
Merge pull request #1399 from martin-frbg/issue1398
Fix LAPACKE build problems with both cmake and make
Martin Kroeker [Thu, 21 Dec 2017 19:42:30 +0000 (20:42 +0100)]
Add conditionals around ar calls for optional modules
The macOS ar aborts when it gets called with no input, see #1398
Martin Kroeker [Thu, 21 Dec 2017 18:43:09 +0000 (19:43 +0100)]
Restore LAPACKE files for Xgeqpf, Xggsvd and Xggsvp
These were inadvertently dropped from the list in my PR #1095
Martin Kroeker [Wed, 13 Dec 2017 19:27:14 +0000 (20:27 +0100)]
Merge pull request #1393 from martin-frbg/daxpybug
Retire Piledriver/Steamroller/Excavator daxpy microkernels as well
Martin Kroeker [Wed, 13 Dec 2017 17:40:39 +0000 (18:40 +0100)]
Retire Piledriver/Steamroller/Excavator daxpy microkernels as well
related to issue #1332
Martin Kroeker [Sun, 10 Dec 2017 20:46:36 +0000 (21:46 +0100)]
Merge pull request #1390 from martin-frbg/daxpybug
Use Sandybridge daxpy kernel on Haswell and Zen for now
Martin Kroeker [Sun, 10 Dec 2017 18:24:31 +0000 (19:24 +0100)]
Use Sandybridge daxpy kernel on Haswell and Zen for now
The testcase from #1332 exposes a problem in daxpy_microk_haswell-2.c that is not seen with
any of the other Intel x86_64 microkernels.
Martin Kroeker [Sat, 9 Dec 2017 21:29:03 +0000 (22:29 +0100)]
Issue1388 (#1389)
* Calculation of chunk range limits was ignoring num_cpu
bug introduced by me in #1262 - should fix #1388
* Calculation of range limits was ignoring num_cpu
bug introduced by me in #1262
* Calculation of chunk range limits was ignoring num_cpu
bug introduced by me in #1262
* Calculation of chunk range limits was ignoring num_cpu
bug introduced by me in #1262
* Calculation of chunk range limits was ignoring num_cpu
bug introduced by me in #1262
* Calculation of chunk range limits was ignoring num_cpu
bug introduced by me in #1262
Andrew [Tue, 5 Dec 2017 18:54:10 +0000 (19:54 +0100)]
warning cleanup (#1380)
* dead increments in driver/level2
* dead increments in kernel/generic
* part dead increments in kernel/x86_64
Martin Kroeker [Tue, 5 Dec 2017 18:53:23 +0000 (19:53 +0100)]
Merge pull request #1382 from martin-frbg/dtrmv-1332
Work around errors in multithreaded dtrmv
Martin Kroeker [Tue, 5 Dec 2017 18:52:52 +0000 (19:52 +0100)]
Merge pull request #1386 from martin-frbg/bignuma
Limit MAX_CPU to 1024 for now
Martin Kroeker [Tue, 5 Dec 2017 18:52:03 +0000 (19:52 +0100)]
Merge pull request #1387 from martin-frbg/cmakeandroid
Explicitly link against libm on Android with cmake as well
Martin Kroeker [Tue, 5 Dec 2017 12:02:48 +0000 (13:02 +0100)]
Explicitly link against libm on Android with cmake as well
Patch from #1384
Martin Kroeker [Tue, 5 Dec 2017 11:54:15 +0000 (12:54 +0100)]
Limit MAX_CPU to 1024 for now
Some Linux distributions (notably SuSE) have raised CPU_SETSIZE to 4096, apparently disregarding API limitations.
From #1348, the highest value to survive array initialization (on a desktop system) is 3232, and 1024 - which is the
more usual CPU_SETSIZE limit, was demonstrated to work fine on an actual bignuma system.
Martin Kroeker [Sun, 3 Dec 2017 21:41:54 +0000 (22:41 +0100)]
Disable gemv unrolling
as a (hopefully temporary) workaround for #1332
Martin Kroeker [Sun, 3 Dec 2017 21:40:54 +0000 (22:40 +0100)]
Disable multithreading for trmv
as a (hopefully temporary) workaround for #1332
Martin Kroeker [Sun, 3 Dec 2017 20:35:20 +0000 (21:35 +0100)]
Merge pull request #1381 from martin-frbg/ctest-warnings
Fix compiler warnings in ctest
Martin Kroeker [Sun, 3 Dec 2017 17:19:30 +0000 (18:19 +0100)]
Fix compiler warnings in ctest
Various fixes for const correctness, stray tab characters and unused labels
Martin Kroeker [Sun, 3 Dec 2017 12:04:02 +0000 (13:04 +0100)]
Merge pull request #1379 from martin-frbg/warnfix
Work around compiler warnings for unused variables
Martin Kroeker [Sat, 2 Dec 2017 21:51:58 +0000 (22:51 +0100)]
Work around compiler warnings for unused variables in the generic zgemm3m_Xcopy kernels
Martin Kroeker [Sat, 2 Dec 2017 15:49:47 +0000 (16:49 +0100)]
Merge pull request #1372 from martin-frbg/param
Correct zgeadd_k prototype
Martin Kroeker [Sat, 2 Dec 2017 11:59:27 +0000 (12:59 +0100)]
Prefix make jobs with travis_wait (#1378)
* Prefix make with travis_wait to prevent it getting killed for producing no output
* Extend travis_wait to 30mins for the windows build
* Trying 45 mins wait time
* Increase travis_wait time to 45 minutes for linux builds as well
Martin Kroeker [Fri, 1 Dec 2017 15:22:35 +0000 (16:22 +0100)]
Merge pull request #1377 from isuruf/threads
Allow overriding NUM_THREADS in cmake
Isuru Fernando [Fri, 1 Dec 2017 07:39:46 +0000 (01:39 -0600)]
Allow overriding NUM_THREADS
Martin Kroeker [Fri, 1 Dec 2017 07:11:12 +0000 (08:11 +0100)]
Merge pull request #1376 from xoviat/patch-2
[appveyor] fix test directory
xoviat [Thu, 30 Nov 2017 22:31:09 +0000 (16:31 -0600)]
[appveyor] fix test directory
Martin Kroeker [Thu, 30 Nov 2017 21:43:54 +0000 (22:43 +0100)]
Merge pull request #1375 from xoviat/patch-1
[appveyor] Use out-of-tree build and cache
xoviat [Thu, 30 Nov 2017 21:33:32 +0000 (15:33 -0600)]
[appveyor] fix syntax
xoviat [Thu, 30 Nov 2017 21:30:10 +0000 (15:30 -0600)]
[appveyor] Use out-of-tree build and cache
Martin Kroeker [Thu, 30 Nov 2017 11:54:52 +0000 (12:54 +0100)]
Merge pull request #1373 from mc10/patch-1
README: Use the SVG Travis badge
Kevin Ji [Wed, 29 Nov 2017 23:21:12 +0000 (15:21 -0800)]
README: Use the SVG Travis badge
Martin Kroeker [Wed, 29 Nov 2017 18:57:35 +0000 (19:57 +0100)]
Correct zgeadd_k prototype
Martin Kroeker [Wed, 29 Nov 2017 18:55:21 +0000 (19:55 +0100)]
Merge pull request #1371 from martin-frbg/develop
Add trivially optimized DSDOT for POWER8
martin [Tue, 28 Nov 2017 17:38:07 +0000 (18:38 +0100)]
Add trivially optimized DSDOT for POWER8
Martin Kroeker [Tue, 28 Nov 2017 17:15:31 +0000 (18:15 +0100)]
Merge pull request #1369 from martin-frbg/dsdot
Add optimized dsdot to all other x86_64 kernels that use sdot.c
Martin Kroeker [Tue, 28 Nov 2017 17:15:04 +0000 (18:15 +0100)]
Merge pull request #1368 from brada4/develop
Eliminate warnings
Martin Kroeker [Sun, 26 Nov 2017 18:12:00 +0000 (19:12 +0100)]
Merge pull request #1366 from martin-frbg/develop
Update LAPACK to 3.8.0
Andrew [Sun, 26 Nov 2017 16:24:08 +0000 (17:24 +0100)]
more dead increments clang4 scan-build deadcode.deadstores
Andrew [Sun, 26 Nov 2017 12:26:11 +0000 (13:26 +0100)]
Eliminate 2-8 dead increments code
Andrew [Sat, 25 Nov 2017 01:54:37 +0000 (02:54 +0100)]
elminate unread variable, after reiteration 3 of them (clang4)
Martin Kroeker [Fri, 24 Nov 2017 19:05:27 +0000 (20:05 +0100)]
Add trivially optimized dsdot based on sdot
Martin Kroeker [Fri, 24 Nov 2017 19:04:29 +0000 (20:04 +0100)]
Add trivially optimized dsdot based on sdot
Martin Kroeker [Fri, 24 Nov 2017 19:03:40 +0000 (20:03 +0100)]
Add trivially optimized dsdot based on sdot
Martin Kroeker [Fri, 24 Nov 2017 19:02:28 +0000 (20:02 +0100)]
Add trivially optimized dsdot based on sdot
Martin Kroeker [Fri, 24 Nov 2017 19:01:42 +0000 (20:01 +0100)]
Add trivially optimized dsdot based on sdot
Martin Kroeker [Fri, 24 Nov 2017 19:00:23 +0000 (20:00 +0100)]
Add trivially optimized dsdot based on sdot
Martin Kroeker [Fri, 24 Nov 2017 18:59:28 +0000 (19:59 +0100)]
Add trivially optimized dsdot based on sdot
Andrew [Fri, 24 Nov 2017 18:13:24 +0000 (19:13 +0100)]
eliminate Wunused-const gcc7 warning
Andrew [Fri, 24 Nov 2017 17:39:04 +0000 (18:39 +0100)]
fix spurious compiler warning fix (no code change)
Andrew [Fri, 24 Nov 2017 17:36:37 +0000 (18:36 +0100)]
fix spurious compiler warning (no code change)
martin [Fri, 24 Nov 2017 08:15:20 +0000 (09:15 +0100)]
fix location of lapacke_nancheck
martin [Fri, 24 Nov 2017 07:15:40 +0000 (08:15 +0100)]
update cmake files
martin [Thu, 23 Nov 2017 20:22:01 +0000 (21:22 +0100)]
update cmakefiles for lapack 3.8.0
projects / platform / upstream / openblas.git / log