floraachy
/
OpenBLAS
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
7,433 Commits 47 Branches 64 Tags 71 MiB
Commit Graph

7433 Commits

Author SHA1 Message Date
Martin Kroeker 0a24f631e9
Merge pull request #3844 from Mousius/switch-ratio-16 
Set SWITCH_RATIO for Arm(R) Neoverse(TM) V1 CPUs
 2022-12-02 12:48:43 +01:00
Martin Kroeker 65984fbe68
Merge pull request #3847 from bartoldeman/scal-benchmark 
scal benchmark: eliminate y, move init/timing out of loop
 2022-12-02 11:51:50 +01:00
Martin Kroeker f6f0d13b9f
Merge pull request #3842 from Mousius/sve-dot 
Add SVE implementation for sdot/ddot
 2022-12-02 08:30:51 +01:00
Bart Oldeman 5c3169ecd8 dscal: use ymm registers in Haswell microkernel 
Using 256-bit registers in dscal makes this microkernel consistent with
cscal and zscal, and generally doubles performance if the vector fits in
L1 cache.
 2022-12-01 07:48:05 -05:00
Chris Sidebottom eea006a688 Wrap SVE header with __has_include check 2022-12-01 12:07:55 +00:00
Chris Sidebottom fd4f52c797 Add SVE implementation for sdot/ddot 
This adds an SVE implementation to sdot/ddot when available, falling back to the previous Advanced SIMD kernel where there's no SVE implementation for the kernel.

All the targets were essentially treating `dot_thunderx2t99.c` as the Advanced SIMD implementation so I've renamed it to better fit with the feature detection.
 2022-12-01 12:07:50 +00:00
Martin Kroeker b6a4ef98b9
Merge pull request #3845 from Mousius/asimd-dot-opt 
Remove unnecessary instructions from Advanced SIMD dot
 2022-11-30 21:07:30 +01:00
Chris Sidebottom 2fb096315e Set SWITCH_RATIO for Arm(R) Neoverse(TM) V1 CPUs 
From testing this yields better results than the default of `2`.
 2022-11-30 09:35:38 +00:00
Bart Oldeman bae45d94d1 scal benchmark: eliminate y, move init/timing out of loop 
Removing y avoids cache effects (if y is the size of the L1 cache, the
main array x is removed from it).
Moving init and timing out of the loop makes the scal benchmark behave like
the gemm benchmark, and allows higher accuracy for smaller test cases since
the loop overhead is much smaller than the timing overhead.

Example:
OPENBLAS_LOOPS=10000 ./dscal.goto 1024 8192 1024
on AMD Zen2 (7532) with 32k (4k doubles) L1 cache per core.

Before
From : 1024  To : 8192 Step = 1024 Inc_x = 1 Inc_y = 1 Loops = 10000
   SIZE       Flops
   1024 :     5627.08 MFlops   0.000000 sec
   2048 :     5907.34 MFlops   0.000000 sec
   3072 :     5553.30 MFlops   0.000001 sec
   4096 :     5446.38 MFlops   0.000001 sec
   5120 :     5504.61 MFlops   0.000001 sec
   6144 :     5501.80 MFlops   0.000001 sec
   7168 :     5547.43 MFlops   0.000001 sec
   8192 :     5548.46 MFlops   0.000001 sec

After
From : 1024  To : 8192 Step = 1024 Inc_x = 1 Inc_y = 1 Loops = 10000
   SIZE       Flops
   1024 :     6310.28 MFlops   0.000000 sec
   2048 :     6396.29 MFlops   0.000000 sec
   3072 :     6439.14 MFlops   0.000000 sec
   4096 :     6327.14 MFlops   0.000001 sec
   5120 :     5628.24 MFlops   0.000001 sec
   6144 :     5616.41 MFlops   0.000001 sec
   7168 :     5553.13 MFlops   0.000001 sec
   8192 :     5600.88 MFlops   0.000001 sec

We can see the L1->L2 switchover point is now where it should be, and the
number of flops for L1 is more accurate.
 2022-11-29 08:02:45 -05:00
lilianhuang fdac8a97c1 Add sbgemm_ncopy_8 and sbgemm_tcopy_4 2022-11-29 04:46:14 -05:00
lilianhuang 135718eafc Improve the performance of sbgemm_tcopy on neoversen2 2022-11-28 04:17:54 -05:00
Chris Sidebottom 4f7b77e08a Remove unnecessary instructions from Advanced SIMD dot 
The existing kernel was issuing extra instructions to organise the arguments into the same registers they would usually be in and similarly to put the result into the appropriate register.

This has an impact on smaller sized dots and seemed like a quick fix
 2022-11-25 16:19:03 +00:00
Martin Kroeker e9a911fb9f
Merge pull request #3841 from martin-frbg/lapack755+764 
Fix SLATRS3 and CLATRS3 tests in TESTING/LIN (Reference-LAPACK PRs 755+764)
 2022-11-23 22:38:06 +01:00
Martin Kroeker bf0e8d67b5
Merge pull request #3840 from martin-frbg/lapack760 
Fix typo in EIG tests and spurious return in lapacke_?tz_trans utility (Reference-LAPACK PR760)
 2022-11-23 19:16:25 +01:00
Martin Kroeker a5470521ee
Fix array indexation in copy, and fix test (Reference-LAPACK PR764) 2022-11-23 15:31:25 +01:00
Martin Kroeker b0393ea4e1
Fix test (Reference-LAPACK PR764) 2022-11-23 15:27:46 +01:00
Martin Kroeker 0d26f1a4c7
Fix wrong indexation in test (Reference-LAPACK PR755) 2022-11-23 15:22:27 +01:00
Martin Kroeker 19fd2d7f00
Use LSAME for character comparison (Reference-LAPACK PR755) 2022-11-23 15:19:07 +01:00
Martin Kroeker 663bf68dbd
Merge pull request #3839 from martin-frbg/lapack758 
Fix array dimesion in complex SYL01 test (Reference-LAPACK PR758)
 2022-11-23 14:57:56 +01:00
Martin Kroeker c2ba4e6249
Remove unnecessary return in void function call (Reference-LAPACK PR760) 2022-11-23 10:43:34 +01:00
Martin Kroeker 74962c7f53
Remove unnecessary return in void function call (Reference-LAPACK PR760) 2022-11-23 10:42:29 +01:00
Martin Kroeker d952cbf7bc
Remove unnecessary return in void function call (Reference-LAPACK PR760) 2022-11-23 10:41:50 +01:00
Martin Kroeker 7694ff495f
Remove unnecessary return in void function call (Reference-LAPACK PR760) 2022-11-23 10:40:59 +01:00
Martin Kroeker 825ae316e2
Fix typo in EXTERNAL (Reference-LAPACK PR760) 2022-11-23 10:36:10 +01:00
Martin Kroeker 730ed549e6
Fix typo in EXTERNAL (Reference-LAPACK PR760) 2022-11-23 10:35:23 +01:00
Martin Kroeker bc3393f703
Fix array dimension (Reference-LAPACK 758) 2022-11-23 10:31:18 +01:00
Martin Kroeker 0b2f8dabbf
Fix array dimension (Reference-LAPACK 758) 2022-11-23 10:30:35 +01:00
Martin Kroeker b4c9228441
Merge pull request #3838 from martin-frbg/lapa311 
Update the version number of the included LAPACK to 3.11.0
 2022-11-22 17:39:51 +01:00
Martin Kroeker e6e2a63650
Update LAPACK version number to 3.11.0 2022-11-22 14:02:21 +01:00
Martin Kroeker 8408357bab
Update LAPACK version number to 3.11.0 2022-11-22 14:01:48 +01:00
Martin Kroeker ba8fb8b4b2
Merge pull request #3837 from martin-frbg/lapack655+697 
Improve convergence of LAPACK ?LAED4 and fix a bug in DORCSD2BY1 (Reference-LAPACK PRs 655+697)
 2022-11-22 13:51:57 +01:00
Martin Kroeker cabf9453e2
Merge pull request #3836 from martin-frbg/lapack665+735 
Fix documentation of LAPACK functions ?TPRFB and IEEECK (Reference-LAPACK PRs 665+735)
 2022-11-22 09:25:24 +01:00
Martin Kroeker d321357558
Fix bug in DORCSD2BY1 (from Reference-LAPACK PR697) 2022-11-21 21:19:44 +01:00
Martin Kroeker afcd7e88b6
Improve convergence of DLAED4/SLAED4 (Reference-LAPACK PR655) 2022-11-21 21:18:39 +01:00
Martin Kroeker f8f2bebf11
Fix function documentation for LAPACK ?TPRFB (Reference-LAPACK PR665) 2022-11-21 20:01:47 +01:00
Martin Kroeker c45edcb537
Fix typo in comment (Reference-LAPACK PR735) 2022-11-21 19:59:33 +01:00
Martin Kroeker 880a3fb20f
Merge pull request #3835 from martin-frbg/lapack217 
Simplify ?SYSWAPR and fix its documentation (Reference-LAPACK 217)
 2022-11-21 19:56:28 +01:00
Martin Kroeker 50aba02910
Simplify ?SYSWAPR and fix its documentation (Reference-LAPACK 217) 2022-11-21 18:00:31 +01:00
Martin Kroeker 0b68dd6a9b
Merge pull request #3834 from martin-frbg/lapack631 
Use new algorithms for computing Givens rotations (Reference-LAPACK PR631)
 2022-11-21 08:30:14 +01:00
Martin Kroeker 9343499256
Merge pull request #3833 from martin-frbg/lapack712+747 
Set scale early in ?LATBS/?LATRS and fix documentation of ?LASCL2 (Reference-LAPACK PRs 712+747)
 2022-11-21 08:29:49 +01:00
Martin Kroeker 7ae4269add
Use new algorithms for computing Givens rotations (Reference-LAPACK PR631) 2022-11-20 22:52:28 +01:00
Martin Kroeker e00f0fb26a
Fix function documentation (Reference-LAPACK PR747) 2022-11-20 22:46:58 +01:00
Martin Kroeker 31d2145988
Set scale early for robust triangular solvers (Reference-LAPACK PR712) 2022-11-20 22:44:36 +01:00
Martin Kroeker 1d5a3aff0d
Merge pull request #3832 from martin-frbg/lapack681+698 
Improve ?LAQR5 and use normwise criterion in ?LAQZ0 (Reference-LAPACK PRs 681+698)
 2022-11-20 22:40:52 +01:00
Martin Kroeker c6816bb576
Use normwise criterion in multishift QZ (Reference-LAPACK PR698) 2022-11-20 19:39:12 +01:00
Martin Kroeker 6f09e4c121
Improve FMA usage in ?LAQR5 (Reference-LAPACK PR681) 2022-11-20 19:37:28 +01:00
Martin Kroeker f63c93274c
Merge pull request #3831 from martin-frbg/lapack647+697+702 
Fix code and documentation for ?SORBDB?/?CUNBDB? (Reference-LAPACK PRs 647+697+702)
 2022-11-20 19:34:41 +01:00
Martin Kroeker aaea0804bc
Fix function documentation (Reference-LAPACK PR697) 2022-11-20 16:38:57 +01:00
Martin Kroeker b946820502
Fix uninitialized variable (Reference-LAPACK PR647) 2022-11-20 16:36:19 +01:00
Martin Kroeker 9e29312c83
Fix type precision and function documentation (Reference-LAPACK PRs 647+702) 2022-11-20 16:34:45 +01:00