43c2e845ab
Switch blas_server to use acq/rel semantics
...
Heavy-weight locking isn't required to pass the work queue
pointer between threads and simple atomic acquire/release
semantics can be used instead. This is especially important as
pthread_mutex_lock() isn't fair.
We've observed substantial variation in runtime because of the
the unfairness of these locks which complety goes away with
this implementation.
The locks themselves are left to provide a portable way for
idling threads to sleep/wakeup after many unsuccessful iterations
waiting.
2020-03-02 02:52:49 +00:00
014fc13995
Merge pull request #2475 from martin-frbg/039changes
...
Update ChangeLog for 0.3.9
2020-03-02 00:04:26 +01:00
f1e05676a0
Merge pull request #2474 from martin-frbg/p9be
...
Use POWER8 kernels on big-endian POWER9 for now
2020-03-02 00:04:08 +01:00
d221c50f27
Add Ampere EMAG8180
2020-03-02 00:02:36 +01:00
f14013da7f
Update with 0.3.9 changes
2020-03-02 00:01:22 +01:00
4f371b0fbf
Use POWER8 kernels on big-endian POWER9 for now
2020-03-01 23:45:58 +01:00
02d60c1563
Merge pull request #35 from xianyi/develop
...
rebase
2020-03-01 23:44:10 +01:00
2e6963259b
Merge pull request #2471 from AGSaidi/l3-fix-2
...
Fix barriers in level3_thread
2020-03-01 19:41:07 +01:00
e94590e400
Merge pull request #2468 from AGSaidi/wfe
...
Use wait-for-event to not spin in the blas_lock
2020-03-01 19:40:46 +01:00
69d4687142
Merge pull request #2464 from Darkness303/develop
...
Add syr benchmark
2020-03-01 13:02:34 +01:00
a731a9bbb9
Merge pull request #2467 from AGSaidi/rpcc
...
Make rpcc() on arm64 get closer to what x86 returns
2020-02-29 22:43:02 +01:00
1aa5907a2c
Merge pull request #2463 from martin-frbg/mingwfix
...
Apply MinGW AVX512 compilation fix to fortran options as well
2020-02-29 19:08:03 +01:00
ea8eec5d17
Merge pull request #2422 from wjc404/develop
...
Adjust SkylakeX GEMM3M parameters, add an AVX512 STRMM kernel and fix performance bugs in AVX2 s/c/z GEMM
2020-02-29 19:07:35 +01:00
97ce6bbce2
Fix barriers in level3_thread
2020-02-29 17:45:17 +00:00
a9aeb6745c
Merge pull request #2465 from AGSaidi/neoverse-n1
...
Add Neoverse-N1 core
2020-02-29 13:24:44 +01:00
0af9991cc9
Use wait-for-event to not spin in the blas_lock
2020-02-29 04:23:48 +00:00
19f3a4091c
Make rpcc() on arm64 get closer to what x86 returns
...
The Arm implementation of rpcc() uses the architected timer
which is defined by the SBSA to be between 10-400MHz. These numbers
are much smaller than the cycle counter frequency used by x86. Make
the numbers closer by shifting the cycle counter up by the number of
leading zeros in the cntfrq_el0 register which gets us closer to a
noraml cpu clock cycle range.
2020-02-29 04:23:22 +00:00
c623a965f9
Add Neoverse-N1 core
...
The implementation is a hybird of the ARMV8 one with some of the
improved TX2 rountines along with specifying -march=v8.2-a
2020-02-29 03:22:04 +00:00
e1062400c4
New add syr benchmark
2020-02-28 16:36:53 +08:00
a66f4d80c8
Apply MinGW AVX512 compilation fix to fortran options as well
...
original issue was #1708 , I see now that the same problem affects gfortran compilation. The underlying issue is said to be fixed (but not yet released) on all branches of gcc as of a few days ago but it will certainly take time to reach mingw/msys.
2020-02-27 23:09:40 +01:00
dd22eb7621
Update cgemm_kernel_8x2_haswell.c
2020-02-27 22:26:15 +08:00
2352331e60
Update zgemm_kernel_4x2_haswell.c
2020-02-27 22:25:19 +08:00
430ee31e66
Merge pull request #2447 from martin-frbg/issue2446
...
Always select ARMV8 parameters for big servers when cpu is TSV110 or EMAG8180
2020-02-27 15:07:02 +01:00
8164fd1328
Always assume server-class cpu count for TSV110 and EMAG8180
2020-02-26 22:19:57 +01:00
531c6b96d6
Merge pull request #34 from xianyi/develop
...
rebase
2020-02-26 22:16:28 +01:00
1b980001dd
Update zgemm_kernel_4x2_haswell.c
2020-02-26 18:38:12 +08:00
2515e1152f
Update cgemm_kernel_8x2_haswell.c
2020-02-26 18:36:54 +08:00
ddcbed6690
Merge pull request #2437 from martin-frbg/issue2434
...
[WIP] Add support for Ampere EMAG8180 ARMV8 cpu
2020-02-25 18:42:52 +01:00
f8ec538c82
Add Ampere EMAG8180
2020-02-25 14:30:00 +01:00
ca4f7dceff
Add parameters for EMAG8180 DYNAMIC_ARCH support with cmake
2020-02-24 20:23:18 +01:00
1ddf9f1067
Add EMAG8180 to arm64 DYNAMIC_ARCH list for cmake
2020-02-24 20:16:18 +01:00
4c5fac5a2b
Typo fix
2020-02-24 20:15:04 +01:00
320e2648cd
Add EMAG8180 to DYNAMIC_CORE list for ARM64
2020-02-24 19:23:46 +01:00
9b732696c6
Add DYNAMIC_ARCH support for ARMV8 EMAG8180
2020-02-24 19:20:00 +01:00
c9dcb3d4a4
Merge pull request #2443 from aaawuanjun/develop
...
[OpenBlas]:benchmark/copy.c has time，x,y data loop problems
2020-02-24 13:14:51 +01:00
3bb7f0138e
Merge pull request #2442 from martin-frbg/lapackpr390
...
Apply fix from Reference-LAPACK PR 390
2020-02-24 12:27:01 +01:00
c93ae92579
[OpenBlas]:benchmark/copy.c has time，x,y data loop problems
2020-02-24 11:23:39 +08:00
87ac1ceb0b
Apply fix from Reference-LAPACK PR390, NaN not propagating
2020-02-23 22:40:40 +01:00
9e40c080f2
Apply fix from Reference-LAPACK PR390, NaN not propagating
2020-02-23 22:39:01 +01:00
903854c168
Add files via upload
2020-02-22 23:40:02 +08:00
a2ff577a30
Update KERNEL.ZEN
2020-02-22 23:39:43 +08:00
97a32cb0a5
Update KERNEL.HASWELL
2020-02-22 23:39:20 +08:00
f1746e7284
Delete sgemm_kernel_8x4_haswell_2.c
2020-02-22 23:38:48 +08:00
f6fcbd7906
Fix performance bug when LDC is a multiple of 1024
2020-02-22 23:37:45 +08:00
1e8410f18c
Merge pull request #2441 from martin-frbg/ismin2
...
Add proper defaults for the IxMIN/IxMAX kernels on mips64 and power
2020-02-22 11:21:03 +01:00
07454bf4d5
Add proper defaults for IxMIN/IxMAX kernels
...
the fallbacks from Makefile.L1 assume a combined source for absolute value and non-absolute (with ifdef USE_ABS) but here we have separate implementations
2020-02-21 11:58:15 +01:00
4046985913
Add proper defaults for IxMIN/IxMAX kernels
...
the fallbacks from Makefile.L1 assume a combined source for absolute value and non-absolute (with ifdef USE_ABS) but here we have separate implementations
2020-02-21 11:55:52 +01:00
75577f95a7
Merge pull request #33 from xianyi/develop
...
rebase
2020-02-21 09:56:05 +01:00
33d92c7a37
Merge pull request #2435 from martin-frbg/issue2433
...
Fix handling of ppc endianness
2020-02-21 00:01:58 +01:00
e57b11acca
Add preliminary support for EMAG8180
2020-02-19 19:00:28 +01:00
Ali Saidi
Martin Kroeker
j00520245
wjc404
wuanjun 00447568