8229c163b7
Use runtime check for AVX512 (sgemm_direct) capability when using DYNAMIC_ARCH
2020-03-26 21:12:56 +01:00
6a14b34c20
Avoid calling DIRECT codepath in DYNAMIC_ARCH on non-SKX
2020-03-22 14:33:16 +01:00
d65e9a2bbd
Merge pull request #2253 from thrasibule/xerbla
...
fix error messages
2020-01-18 20:39:04 +01:00
2463938879
fix error message
2019-09-11 10:35:25 -04:00
5d6525c87c
more bugfix
2019-09-10 17:30:57 -04:00
459bb9291d
fix error codes
2019-09-10 17:10:33 -04:00
5997b6b491
bugfix
2019-09-08 11:14:49 -04:00
7ec7b999a5
add missing file
2019-09-08 11:14:49 -04:00
af9ac0898a
fix Makefile
2019-09-08 11:14:49 -04:00
9b2f0323d6
update Makefile
2019-09-08 11:14:49 -04:00
ea747cf933
start working on ?trtrs
2019-09-08 11:14:49 -04:00
daf2fec12d
Misc. typo fixes
...
Found via `codespell -q 3 -w -L ith,als,dum,nd,amin,nto,wis,ba -S ./relapack,./kernel,./lapack-netlib`
2019-04-29 17:03:56 -04:00
268c28db7d
Merge pull request #2095 from martin-frbg/trsm
...
Correct length of name string in xerbla call
2019-04-28 09:55:25 +02:00
0bd956fd21
Correct length of name string in xerbla call
2019-04-27 22:49:04 +02:00
79cfc24a62
Add interface for ?sum (derived from ?asum)
2019-03-30 21:59:18 +01:00
c19a449096
Merge pull request #2071 from martin-frbg/issue2068
...
Provide CBLAS interfaces to I?MIN and I?MAX
2019-03-30 14:54:28 +01:00
3d1e36d4cb
Build CBLAS interfaces for I?MIN and I?MAX
2019-03-30 12:38:41 +01:00
e29b0cfcc4
Allow multithreading TRMV again
...
revert workaround introduced for issue #1332 as the actual cause appears to be my incorrect fix from #1262 (see #1388 )
2019-02-19 21:03:30 +01:00
8533aca964
Avoid penalizing tall skinny matrices
2019-01-23 10:03:00 +01:00
cda81cfae0
Shift transition to multithreading towards larger matrix sizes
...
See #1886 and JuliaRobotics issue 500. trsm benchmarks on Haswell and Zen showed that with these values performance is roughly doubled for matrix sizes between 8x8 and 14x14, and still 10 to 20 percent better near the new cutoff at 32x32.
2019-01-19 00:10:01 +01:00
cdc668d82b
Add a "sgemm direct" mode for small matrixes
...
OpenBLAS has a fancy algorithm for copying the input data while laying
it out in a more CPU friendly memory layout.
This is great for large matrixes; the cost of the copy is easily
ammortized by the gains from the better memory layout.
But for small matrixes (on CPUs that can do efficient unaligned loads) this
copy can be a net loss.
This patch adds (for SKYLAKEX initially) a "sgemm direct" mode, that bypasses
the whole copy machinary for ALPHA=1/BETA=0/... standard arguments,
for small matrixes only.
What is small? For the non-threaded case this has been measured to be
in the M*N*K = 28 * 512 * 512 range, while in the threaded case it's
less, around M*N*K = 1 * 512 * 512
2018-12-13 13:47:31 +00:00
5393759a98
Merge pull request #1869 from martin-frbg/axpy0
...
Handle special case INCX=0,INCY=0 in the axpy interface
2018-11-25 20:52:49 +01:00
c171b8ad13
Handle special case INCX=0,INCY=0 in the axpy interface
2018-11-13 13:57:18 +01:00
96d2f2c9b2
Merge pull request #1831 from brada4/hemv
...
disable threading in C/ZSWAP copying from S/DSWAP
2018-11-07 08:49:21 +01:00
2992e3886a
disable threading in C/ZSWAP copying from S/DSWAP
2018-10-22 23:21:49 +03:00
e3c262e5cf
Merge pull request #1825 from brada4/hemv
...
Delay _hemv threading in attempt to address #1820
2018-10-21 20:34:05 +02:00
a293bdcd5e
re-arrange new code for readability
2018-10-20 21:37:53 +03:00
c7bbf9c987
Attempt to tame _hemv threading #1820
2018-10-20 11:13:29 +03:00
21f46a1cf2
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.
2018-10-17 10:44:37 -07:00
b991570210
Merge pull request #1762 from martin-frbg/issue1710-2
...
Add explicit casts to silence compiler warnings
2018-09-19 18:16:21 +02:00
f3c262156e
Add an explicit cast to silence a warning
...
for #1710
2018-09-13 14:24:29 +02:00
30f5a69ab8
Add explicit cast to silence a warning
...
for #1710
2018-09-13 14:23:31 +02:00
4a553e8678
Merge pull request #1713 from martin-frbg/issue1710
...
Introduce blasabs macro and use it to switch between abs and labs for INTERFACE64
2018-08-04 23:51:31 +02:00
165f00c159
fabs -> fabsl
2018-08-04 20:14:51 +02:00
933896a1d0
Use blasabs to switch between abs and labs as needed for INTERFACE64
2018-08-04 20:06:49 +02:00
a4e321400b
fabs -> fabsl
...
Fixes two calls that were using `fabs` on a `long double` argument rather than `fabsl`, which looks like it is doing an unintentional truncation to `double` precision.
2018-08-03 13:00:10 -04:00
9cf22b7d91
Build cblas_iXamin interfaces
2018-06-23 13:27:30 +02:00
c2545b0fd6
Fixed a few more unnecessary calls to num_cpu_avail.
...
I don't have as many benchmarks for these as for gemm, but it should still
make a difference for small matrices.
2018-06-11 10:17:16 +01:00
66316b9f4c
Improve performance of GEMM for small matrices when SMP is defined.
...
Always checking num_cpu_avail() regardless of whether threading will actually
be used adds noticeable overhead for small matrices. Most other uses of
num_cpu_avail() do so only if threading will be used, so do the same here.
2018-06-07 15:29:13 +01:00
e8880c1699
Use a single thread for small input size
...
copies daxpy improvement from #27 , see #1560
2018-06-07 10:26:55 +02:00
1d27fa8507
Merge pull request #1539 from martin-frbg/ztrmv-1332
...
Disable multithreading in ztrmv
2018-04-27 23:10:21 +02:00
a8ed428bab
Disable multithreading in ztrmv
...
BLAS-Tester shows that the same problem exists as with DTRMV (issue #1332 )
2018-04-25 22:35:46 +02:00
809fd0d451
Rewrite ROTMG to address cases not covered by the netlib algorithm ( #1480 )
...
* Rewrite ROTMG based on the new implementation in GONUM based on the algorithm proposed by Tim Hopkins, see issue 1452 for the reference
* Correct ROTMG utest for issue1452 and add another from gonum, also correct transposition of expected and observed values in error messages
2018-03-04 17:39:56 +01:00
72f14a0363
Fix conditionals in the rescaling against GAMSQ
2018-02-18 12:54:52 +01:00
798f1595d5
Fix condition in both second scaling loops
2018-02-18 12:37:09 +01:00
0464aa6784
Remove debug printfs
2018-02-09 23:06:50 +01:00
55840f0bc9
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.
2018-02-09 23:00:03 +01:00
47deec2c1a
fix couple of dead assignment warnings
2017-12-22 00:56:35 +01:00
38763ec4f3
Disable multithreading for trmv
...
as a (hopefully temporary) workaround for #1332
2017-12-03 22:40:54 +01:00
9251a2efde
Merge pull request #1359 from brada4/develop
...
Eliminate mode variable where not needed in syrk interface
2017-11-18 23:47:17 +01:00
b46e2b57cc
Make return parameter of cblas_Xdotc_sub, cblas_Xdotu_sub a void pointer as well
2017-11-18 20:28:02 +01:00
3ce401f51b
Make last parameter of cblas_Xdotc_sub/cblas_Xdotu_sub a void pointer as well
2017-11-18 18:58:40 +01:00
27575d200a
Eliminate mode variable where not needed
2017-11-15 15:32:38 +01:00
2c222f1faa
Modify complex CBLAS functions to take void pointers
...
Modify complex CBLAS functions to take void pointers instead of float or double arguments (to bring the prototypes in line with netlib and other implementations' cblas.h)
2017-11-05 15:53:14 +01:00
742f54c235
Merge pull request #1303 from martin-frbg/imatcopy-rowscols
...
Fix cols/rows mixup in omatcopy 2nd step for BlasTrans cases
2017-09-14 21:46:26 +02:00
d674fbb4c7
Fix cols/rows mixup in omatcopy 2nd step for BlasTrans cases
...
Equivalent of #1244 (issue #899 ) for the non-complex cases. Fixes #1289
2017-09-14 19:59:05 +02:00
46c9357c72
Merge pull request #1288 from quickwritereader/develop
...
Optimized standard Blas Level-1,2 (excluding nrm2 functions) for z13 (double precision). Issue 884
2017-09-09 23:47:17 +02:00
1cfdb2295d
Optimized standard Blas Level-1,2 (excluding nrm2 functions) for z13 (double precision)
2017-09-06 16:41:08 +04:00
00740c0e34
Merge pull request #1290 from martin-frbg/imatcopy
...
Use in-place transform shortcut only if matrix is square
2017-09-03 13:02:10 +02:00
254db9bd7c
Use in-place transform shortcut only if matrix is square
2017-09-03 09:52:55 +02:00
d245caa49a
Support out-of-source build
2017-08-01 15:16:14 +05:30
376048156b
Use in-place transform shortcut only if matrix is square
2017-07-21 11:20:15 +02:00
d1c5b8f913
Add files via upload
2017-07-20 20:51:06 +02:00
91bde7d315
Exchange rows and cols in final omatcopy with BlasTrans
...
This is MicMuc's patch from #899
2017-07-15 22:02:53 +02:00
1e06b49854
Update xerbla.c
2017-04-26 20:29:30 +02:00
7f546f54fa
Add cblas_xerbla
2017-04-26 20:01:34 +02:00
a809431e34
Add cblas_xerbla()
2017-04-26 19:58:59 +02:00
99880f7906
Address unlikely memleak in zimatcopy interface ( #1129 )
...
* fix unlikely memleak in zimatcopy interface
* fix only unlikely memleak in zimatcopy interface
* fix only unlikely memleak in zimatcopy interface
2017-03-16 13:13:31 +01:00
211d2eceb5
Update zdot.c
2017-03-13 18:08:00 +01:00
5813ed095b
Update zdot.c
2017-03-13 17:49:07 +01:00
e44b028fe5
Replace gnu _real_, _imag_ extensions in initializers
2017-03-13 00:40:11 +01:00
071a830e8b
THUNDERX2T99: Add optimized S/D/C/Z SWAP Implementations
2017-02-03 03:55:06 -08:00
dd6212e684
updated some level1 funcions, that are not thread save
2017-01-10 14:05:07 +01:00
84b8170bfb
Adding multi-threading for copy, dot, rot, and asum funcitons
2017-01-10 11:48:58 +08:00
ae4ac6f984
removed obj-files, that are moved to lapack 3.7.0
2017-01-06 16:14:53 +01:00
d346c533b1
Fix z/ctrmv stack allocation on AMD bulldozer and barcelona target
...
* Hopefully, because this was found by error and trial (dark magic)
* Ref #786
2016-06-07 16:11:09 +02:00
f04af36ad0
Merge pull request #898 from wernsaar/develop
...
added experimental support for optimized lapack fortran functions
2016-05-31 14:13:52 +02:00
41000c8443
added directory for optimized lapack fortan codes and added dlaqr5.f
2016-05-31 12:53:07 +02:00
053044ae4d
Replace CMAKE_SOURCE_DIR/CMAKE_BINARY_DIR with PROJECT_SOURCE_DIR/PROJECT_BINARY_DIR
...
If OpenBLAS is built using add_subdirectory(OpenBlas) as part of another project
then the paths set by CMAKE_XXX_DIR are relative to the parent project
and not the OpenBLAS project.
2016-05-25 09:13:28 +02:00
40af513669
Disable multi-threading in swap
...
* Close #873
2016-05-16 13:07:55 +00:00
16ec5323c9
Fix zgemv.c compilation when stack allocation is disabled
2016-02-08 12:05:02 +01:00
5fc2203d8a
zgemv: Add a workaround for #746
2016-02-08 11:25:15 +01:00
78dcf5c3d5
Improve performances of ztrmv on small matrices
...
* Use stack allocation
* Disable multi-threading
* Ref #727
2016-02-08 11:25:02 +01:00
32f793195f
Use stack allocation in zgemv and zger
...
For better performance with small matrices
Ref #727
2016-02-08 11:24:21 +01:00
1fe3aab047
Use GEMM_MULTITHREAD_THRESHOLD as a number of ops
...
...not a matrix size. For GEMM_MULTITHREAD_THRESHOLD=4
(the default value) this does not change anything but
for other values it make the GEMM and GEMV thresholds
changing in the same way.
Close #742
2016-01-24 11:31:40 +01:00
1a1935507b
[z]ger: increase multithread threshold
...
The ones given in 3ae30cd
2016-01-24 10:46:35 +01:00
66eafb16cf
swap: disable multi-threading for small matrices
...
Close #731
2016-01-19 17:14:46 +01:00
3ae30cd6b9
Disable multi-threading for small matrices in [z]ger
...
Ref #731
2016-01-19 17:14:31 +01:00
87a2ccc37c
Factorize MAX_STACK_ALLOC code to common_stackalloc.h
...
Ref #727
2016-01-08 16:03:52 +01:00
f9890a6452
Fix compilation when MAX_STACK_ALLOC is not set
...
Close #722
2015-12-31 14:43:09 +01:00
285d042b10
Fixed rotg bug on ARM.
2015-12-14 10:07:01 -06:00
640cccc2b1
Refs #697 . Fixed gemv bug for Windows.
...
Thank matzeri's patch.
2015-11-30 15:19:45 -06:00
55a0b27c01
Minor C code fixes in interface/
2015-11-09 14:15:49 +05:30
2feef49fa8
Merge branch 'develop' into cmake
...
Conflicts:
driver/others/memory.c
2015-10-26 14:54:34 -05:00
5a291606ad
Refs #671 . the return of i?max cannot larger than N.
2015-10-24 01:16:34 +08:00
8fade093aa
Fixed cmake bug on Visual Studio.
2015-10-20 14:37:22 -05:00
94b125255f
Merge branch 'develop' into cmake
...
Conflicts:
driver/others/memory.c
2015-10-13 04:46:08 +08:00
baec8f5cac
Refs #638 . Fixed compiling bug with clang on Mac OS X.
2015-09-10 10:32:07 -05:00
711ca33bc6
Improved Ximatcopy when lda==ldb.
...
The Ximatcopy functions create a copy of the input matrix
although they seem to work inplace. The new routines
XIMATCOPY_K_YY perform the operations inplace if the leading
dimension does not change.
2015-09-07 14:36:16 +02:00
f874465bb8
Use cmake to build OpenBLAS GENERIC Target on MSVC x86 64-bit.
...
Disable CBLAS and LAPACK.
2015-08-10 14:10:44 -05:00
dcd5ba4443
Merge branch 'cmake' of https://github.com/hpanderson/OpenBLAS into hpanderson_cmake
2015-07-22 04:06:39 +08:00
f8f2e261fe
use only 1 thread if m or n < 2*GEMM_MULTITHREAD_THRESHOLD
2015-05-06 10:41:53 +02:00
ab567d8443
gemv: Ensure stack buffer is large enough to handle memory alignment
...
Ref #478
2015-04-24 10:12:49 +02:00
847e19c04e
Refs #478,#482, Enable stack alloc for s/dgemv_t.(revert 9798491)
2015-04-20 23:22:40 -05:00
fd9fd42936
Refs #478 , #482 . Fixed bug on previous commit.
2015-04-13 23:22:27 -05:00
9798481979
Refs #478 , #482 . Fix segfault bug for gemv_t with MAX_ALLOC_STACK flag.
...
For gemv_t, directly use malloc to create the buffer.
2015-04-13 19:45:27 -05:00
cdefdb21cd
Refs #492 . Fixed c/zsyr bug with negative incx.
2015-02-26 06:37:03 +08:00
0d8e227ea7
Changed strategy for setting preprocessor definitions.
...
Instead of generating separate object files for each permutation of
defines for a source file, GenerateNamedObjects now writes an entirely
new source file and inserts the defines as #define c statements.
This solves a problem I ran into with ar.exe where it was refusing to
link objects that had the same filename despite having different paths.
2015-02-24 12:26:33 -06:00
b2284647a3
More complex objects.
2015-02-23 07:51:05 -06:00
a6116e5859
Added some more complex-only objects.
2015-02-22 17:49:28 -06:00
67e39bd8fb
Added mangled complex filenames to interface and lapack CMakeLists.txt.
2015-02-17 13:12:30 -06:00
9eb1499095
Added another param to GenerateNamedObjects to mangle complex source names.
...
There are a lot of sources for complex float types that are the same
names as the real sources, except with z prepended.
2015-02-17 10:30:28 -06:00
39cc6b21d3
Add ATLAS-style ?geadd function
2015-02-16 13:46:20 +01:00
4662a0b13a
Changed generate functions to iterate through a list of float types.
...
This will generate obj files for SINGLE/DOUBLE/COMPLEX/DOUBLE COMPLEX.
2015-02-15 17:44:37 -06:00
e74462a3f5
Moved declarations to start of functions to satisfy MSVC C89 implementation.
2015-02-11 11:16:57 -06:00
e8c39138c6
Removed return value from GenerateNamedObjects.
...
It sets DBLAS_OBJS directly to save a bunch of list appending in the
CMakeLists.txt files.
2015-02-09 12:28:09 -06:00
58cff2fed8
Added CBLAS define/naming convention to GenerateNamedObjects.
2015-02-04 11:30:15 -06:00
5690cf3f0e
Added override for function names in GenerateNamedObjects.
...
The BLAS interface folder should now be generated the correct objects
for the DOUBLE case.
2015-02-04 10:52:19 -06:00
a0aeda6187
Added function to set defines for the object names (e.g. -DNAME=dgemm).
2015-02-04 10:37:34 -06:00
20e593a44a
Added cblas_ objects to interface CMakeLists.
...
Naming isn't right, though, not seeing cblas_xxxx exports in the
resulting library.
2015-02-02 16:25:30 -06:00
9e154aba58
Added LAPACK object files to interface CMakeLists.
2015-02-02 12:31:15 -06:00
5057a4b4df
Added openblas add_library call that uses DBLAS_OBJS ojbects.
2015-01-30 15:21:21 -06:00
a6cf8aafc0
Updated level3/CMakeLists with correct defines using all combos.
2015-01-30 11:21:50 -06:00
b17ccb4c5c
Fix a segfault in gemv when MAX_STACK_ALLOC is set
...
* stack_alloc_size is needed after the implementation call
but it may be overwritten if it's optimized to a register,
because some gemv implementation (ex: dgemv_n.S) do not
restore all register (ex: r10).
* do the same in ger.c for the same reasons even if the bug
has not been observed.
2015-01-29 09:55:57 +01:00
5eefe18ae4
Added CMakeLists.txt for the first of the BLAS folders.
...
It only does the double precision compile currently.
I realized I didn't finish converting Makefile.system yet, so I made
a note of that.
2015-01-27 16:17:17 -06:00
e9d9a8eae3
Allow to do gemv and ger buffer allocation on the stack
...
ger and gemv call blas_memory_alloc/free which in their turn
call blas_lock. blas_lock create thread contention when matrices
are small and the number of thread is high enough. We avoid
call blas_memory_alloc by replacing it with stack allocation.
This can be enabled with:
make -DMAX_STACK_ALLOC=2048
The given size (in byte) must be high enough to avoid thread contention
and small enough to avoid stack overflow.
Fix #478
2014-12-27 14:33:12 +01:00
9e829ce98f
enabled cblas gemm3m functions
2014-09-20 17:20:02 +02:00
d49fd33885
disabled SYMM3M and HEMM3M functions because segment violations
2014-09-20 15:27:40 +02:00
7aae4a62e7
enabled use of GEMM3M functions
2014-09-20 14:27:10 +02:00
3300f5ebff
optimized multithreading lower limits
2014-09-15 11:38:25 +02:00
fd2478c9e2
optimized interface/zgemv.c for multithreading
2014-09-12 19:18:23 +02:00
1cba8e7b11
Merge pull request #446 from grisuthedragon/cblas_matcopy
...
Add a CBLAS interface for the BLAS extension s/d/c/z*matcopy routines.
2014-09-10 16:31:31 +08:00
a057e5434d
add CBLAS interface for s/d/c/zimatcopy
2014-09-09 09:52:13 +02:00
7794766d3c
Add cblas_(s/d/c/z)omatcopy in order to have cblas interface for them.
2014-09-08 17:57:44 +02:00
f511807fc0
modified multithreading threshold
2014-09-08 12:27:32 +02:00
d1800397f5
optimized interface/gemv.c for multithreading
2014-09-02 17:36:07 +02:00
f4ff889491
updated interface/gemv.c for multithreading
2014-09-02 16:30:04 +02:00
51413925bd
adjust number of threads for small size in cgemv and zgemv
2014-07-15 16:27:02 +02:00
b985cea65d
adjust number of threads for sgemv and dgemv
2014-07-15 16:04:46 +02:00
d286daa2ba
adjusted number of threads for small size
2014-07-15 14:41:35 +02:00
cedc1f4b14
Ref #410 : disabled optimized potri functions ( single threading bug)
2014-07-10 13:42:32 +02:00
02a504c0b8
fixed my bug in ger.c
2014-07-02 10:39:33 +02:00
be94db096c
disabled *3M functions for x86_64 platforms
2014-07-01 16:18:05 +02:00
aee61456a4
disabled SMP for sbmv and zsbmv again
2014-06-29 21:18:38 +02:00
01a119abfc
enabled SMP for sbmv and zsbmv, but only for 64bit binaries
2014-06-29 20:35:56 +02:00
1fad2b759f
enabled smp for ger.c and zger.c, but only for 64bit binaries
2014-06-29 16:43:04 +02:00
6c2ead30f0
Remove all trailing whitespace except lapack-netlib
...
Signed-off-by: Timothy Gu <timothygu99@gmail.com>
2014-06-27 12:05:18 -07:00
15d5dfa92c
fixed compiler warnings
2014-06-25 11:32:44 +02:00
86d8c8978b
Ref #391 : disabled SMP in ger.c and zger.c
2014-06-22 12:01:24 +02:00
a19d209005
Ref #103 : enhancement for small matrix dimensions
2014-06-18 15:04:11 +02:00
Martin Kroeker
Guillaume Horel
luz.paz
Arjan van de Ven
Andrew
Ashwin Sekhar T K
Steven G. Johnson
Craig Donner
Andrew
Abdurrauf
Isuru Fernando
Werner Saar
jiahaipeng
Jerome Robert
John Biddiscombe
Zhang Xianyi
Ralph Campbell
Martin Koehler
Hank Anderson
Jerome Robert
Martin Koehler
Martin Köhler
Timothy Gu