floraachy/OpenBLAS
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
6,242 Commits 47 Branches 64 Tags
2e2c02b762afd67fe3cfb49620ab9df721f1a8ea
Commit Graph

41 Commits

Author SHA1 Message Date
H.J. Lu
53ee0b76bb x86: Enable Intel CET 
When Intel CET is enabled, we need to include <cet.h> in assembly codes
to mark Intel CET support and place _CET_ENDBR at the function entry.
 2021-04-30 19:45:39 -07:00
Chen, Guobing
deaeb6c5b8 Add bfloat16 based dot and conversion with single/double 
1. Added bfloat16 based dot as new API: shdot
2. Implemented generic kernel and cooperlake-specific (AVX512-BF16) kernel for shdot
3. Added 4 conversion APIs for bfloat16 data type <=> single/double: shstobf16 shdtobf16 sbf16tos dbf16tod
     shstobf16 -- convert single float array to bfloat16 array
     shdtobf16 -- convert double float array to bfloat16 array
     sbf16tos  -- convert bfloat16 array to single float array
     dbf16tod  -- convert bfloat16 array to double float array
4. Implemented generic kernels for all 4 conversion APIs, and cooperlake-specific kernel for shstobf16 and shdtobf16
5. Update level1 thread facilitate functions and macros to support multi-threading for these new APIs
6. Fix Cooperlake platform detection/specify issue when under dynamic-arch building
7. Change the typedef of bfloat16 from unsigned short to more strict uint16_t

Signed-off-by: Chen, Guobing <guobing.chen@intel.com>
 2020-09-04 02:31:25 +08:00
Martin Kroeker
6c33764ca4 Unify BUFFER_SIZE settings for x86_64 again to fix potentially fatal mismatch in DYNAMIC_ARCH builds 2020-07-22 17:30:55 +00:00
Martin Kroeker
0464e662ad make blas_quickdivide unsigned and guard against miscompilation 2020-06-05 10:03:36 +02:00
Martin Kroeker
a52bdd9d7b Add (empty) read barrier definition 2020-04-13 12:22:35 +02:00
Martin Kroeker
a33d177430 Increase default BUFFER_SIZE on ARM, ZARCH and newer x86_64, add GEMM_R for POWER8/9 
As shown in #2538, default buffersizes on some platforms were smaller than required in memory.c
and the requirement could never be fulfilled for a calculated GEMM_R on PPC given the fomula used
 2020-04-12 19:44:48 +02:00
Martin Kroeker
c353d8b106 Make BUFFER_SIZE configurable 2020-02-09 23:30:22 +01:00
Martin Kroeker
280552b988 Fix mov syntax 2019-06-16 18:35:43 +02:00
Martin Kroeker
bbd4bb0154 Zero ecx with a mov instruction 
PGI assembler does not like the initialization in the constraints.
 2019-06-16 15:04:10 +02:00
luz.paz
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
Martin Kroeker
b55c586fac Fix missing clobber in x86/x86_64 blas_quickdivide inline assembly function (#2017) 
* Fix missing clobber in blas_quickdivide assembly
 2019-02-14 15:21:36 +01:00
Martin Kroeker
0afaae4b23 Query AVX2 and AVX512VL capability in x86 cpu detection 2019-01-05 16:58:56 +01:00
Arjan van de Ven
2ddc96c9e5 make WMB / MB safer on x86-64 
make it so that

if (foo)
	RMB;
else
	MB;

is always done correctly and without syntax surprises
 2018-06-17 18:06:24 +00:00
Arjan van de Ven
7e39ffe113 On x86-64, make MB/WMB compiler barriers 
Whie on x86(64) one does not normally need full memory barriers, it's
good practice to at least use compiler barriers for places where on other
architectures memory barriers are used; this prevents the compiler
from over-optimizing.
 2018-06-17 17:53:15 +00:00
Martin Kroeker
88e224f4c0 Merge pull request #1542 from martin-frbg/quickdiv64 
Avoid out-of-bounds accesses in blas_quickdivide on big X86 systems
 2018-05-02 18:11:50 +02:00
Martin Kroeker
d0c0506588 Omit the divide table overflow check on small systems 2018-05-02 14:44:50 +02:00
Martin Kroeker
c1eb06e102 Update common_x86_64.h 2018-04-29 14:40:12 +02:00
Martin Kroeker
26ce518d46 Avoid out of bounds reads from blas_quick_divide_table on big systems 
Should fix #1541
 2018-04-29 14:34:33 +02:00
Alex Arslan
a41d241a0e Add support for DragonFly BSD 2018-04-03 16:39:29 -07:00
Alex Arslan
8da6b6ae52 Allow building on OpenBSD 
With this change, OpenBLAS builds and all tests pass on OpenBSD 6.2
using Clang. Tested on x86-64 only, with and without DYNAMIC_ARCH=1.
 2018-04-02 10:48:22 -07:00
Paul Osmialowski
d7afdf9137 build: Flang has the same interface as PGI 
Signed-off-by: Paul Osmialowski <pawel.osmialowski@arm.com>
 2017-05-27 06:26:48 +01:00
Keno Fischer
d5e1255ca7 Don't pass REALNAME to .end 
Putting the procedure there is an MSVC-ism, where it is optional. GCC silently ignores and Clang errors, so it is best to remove this.
 2016-03-13 18:56:21 -04:00
Zhang Xianyi
94b125255f Merge branch 'develop' into cmake 
Conflicts:
	driver/others/memory.c
 2015-10-13 04:46:08 +08:00
Grazvydas Ignotas
6b92204a7c add fallback blas_lock implementation 
to be used on armv5 and new platforms
 2015-08-16 18:59:17 +02:00
Grazvydas Ignotas
e12cf1123e add fallback rpcc implementation 
- use on arm, arm64 and any new platform
- use faster integer math instead of double
- use similar scale as rdtsc so that timeouts work
 2015-08-16 18:59:16 +02:00
Zhang Xianyi
f8eba3d548 Fixed cmake build bugs on Linux. 2015-08-11 16:25:16 -05:00
Zhang Xianyi
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
Zhang Xianyi
51ff17d46e Add AMD Excavator target. 2015-05-13 16:16:30 -05:00
Werner Saar
4319769b79 added target processor STEAMROLLER 2014-12-28 20:16:46 +08:00
wernsaar
7794237475 undef WHEREAMI 2014-09-06 11:01:42 +02:00
wernsaar
2021d0f9d6 experimentally removed expensive function calls 2014-09-05 15:05:53 +02:00
Timothy Gu
6c2ead30f0 Remove all trailing whitespace except lapack-netlib 
Signed-off-by: Timothy Gu <timothygu99@gmail.com>
 2014-06-27 12:05:18 -07:00
Zhang Xianyi
16eb780e13 Refs #262. Fixed compatibility issues of GNU stack markings with PathScale EKOPath(tm) Compiler Suite: Version 4.0.12.1 2013-09-22 09:37:59 +08:00
Zhang Xianyi
a2930664f4 Refs #262. Added executable stack markings. 2013-07-28 00:09:40 +08:00
Zhang Xianyi
886cbaf4e4 Support AMD Piledriver by bulldozer kernels. 2013-07-06 12:06:43 -03:00
Zhang Xianyi
88c272f6a7 Refs #83. Added the missing ALIGN_5 macro on Mac OSX. However, it still exists SEGFAULT bug. 2012-06-20 09:20:20 +08:00
Zhang Xianyi
37edae1c90 Refs #75. Check ffreep macro before the define. 2012-05-31 17:17:02 +08:00
Xianyi Zhang
a4daa34db7 Refs #75. Use ffreep opcode directly. Please check out http://www.sandpile.org/x86/opc_fpu.htm . 2012-05-30 20:25:01 +08:00
Zaheer Chothia
a431042475 Fix inconsistent case for OS_* macros (Refs pull request #111) 2012-05-23 00:01:14 +02:00
Mike Nolta
4e29b6ffc0 FreeBSD: fix OS_FreeBSD -> OS_FREEBSD typos 2012-05-21 16:57:19 -04:00
Xianyi Zhang
342bbc3871 Import GotoBLAS2 1.13 BSD version codes. 2011-01-24 14:54:24 +00:00