Chip Kerchner
9ac0fb0111
Merge branch 'develop' into vectorizeBF16GEMV
2024-10-04 06:49:53 -05:00
Martin Kroeker
d714013ab9
change sgemm kernel to 4x4 as the 16x4 altivec goes out of bounds
2024-10-03 22:04:20 +02:00
Chip Kerchner
915a6d6e44
Add casting.
2024-10-03 14:08:21 -05:00
Chip Kerchner
7ec3c16d82
Remove beta from optimized functions.
2024-10-03 13:27:33 -05:00
Martin Kroeker
de421b7764
Merge pull request #4904 from XiWeiGu/la64_cross_cmake
...
LoongArch64: Enable cmake cross-compilation
2024-10-03 15:53:57 +02:00
Chip Kerchner
7cc00f68c9
Remove more duplicate.
2024-10-01 11:23:32 -05:00
Chip Kerchner
e238a68c03
Remove duplicate.
2024-10-01 11:06:23 -05:00
Chip Kerchner
32095b0cbb
Remove parameter.
2024-10-01 09:32:42 -05:00
gxw
30af9278dc
LoongArch64: Enable cmake cross-compilation
2024-09-29 10:13:30 +08:00
gxw
48698b2b1d
LoongArch64: Rename core
...
Use microarchitecture name instead of meaningless strings to name the core,
the legacy core is still retained.
1. Rename LOONGSONGENERIC to LA64_GENERIC
2. Rename LOONGSON3R5 to LA464
3. Rename LOONGSON2K1000 to LA264
2024-09-29 09:35:21 +08:00
Chip Kerchner
c8788208c8
Fixing block issue with transpose version.
2024-09-27 13:27:03 -05:00
Chip Kerchner
d7c0d87cd1
Small changes.
2024-09-26 15:21:29 -05:00
Chip Kerchner
eb6f3a05ef
Common MMA code.
2024-09-26 09:28:56 -05:00
Chip Kerchner
fb287d17fc
Common code.
2024-09-25 16:31:36 -05:00
Chip Kerchner
8ab6245771
Small change.
2024-09-24 16:50:21 -05:00
Chip Kerchner
df19375560
Almost final code for MMA.
2024-09-24 16:30:01 -05:00
Chip Kerchner
05aa63e738
More MMA BF16 GEMV code.
2024-09-24 12:54:02 -05:00
Chip Kerchner
c9ce37d527
Force vector pairs in clang.
2024-09-23 08:43:58 -05:00
Chip Kerchner
89a12fa083
MMA BF16 GEMV code.
2024-09-23 06:32:14 -05:00
Chip Kerchner
7947970f9d
Move common code.
2024-09-13 06:22:13 -05:00
Chip Kerchner
72216d28c2
Fix bug with inc_y adding results twice.
2024-09-11 08:47:32 -05:00
Chip Kerchner
2f142ee857
More common code.
2024-09-09 14:41:55 -05:00
Chip Kerchner
39fd29f1de
Minor improvement and turn off BF16 GEMV forwarding by default.
2024-09-08 18:28:31 -05:00
Chip Kerchner
8541b25e1d
Special case beta is one.
2024-09-06 14:48:48 -05:00
Chip Kerchner
76227e2948
Initial commit for vectorized BF16 GEMV. Added GEMM_GEMV_FORWARD_BF16 to enable using BF16 GEMV for one dimension matrices. Updated unit test to support inc_x != 1 or inc_y for GEMV.
2024-09-06 14:03:31 -05:00
Deeksha Goplani
4894c54055
Improve TN case with further unrolling
2024-09-02 22:22:49 +05:30
Martin Kroeker
e05d98d00a
expressly use fld.d/fst.d for floating point registers instead of LD/ST macros
2024-08-15 22:14:29 +02:00
Chip Kerchner
a0aeba631d
Merge branch 'develop' into betterPowerGEMVTail
2024-08-15 08:00:00 -05:00
Chip Kerchner
083faf7556
Merge branch 'develop' into betterPowerGEMVTail
2024-08-14 15:56:03 -05:00
Chip Kerchner
75472b830a
Merge branch 'develop' into betterPowerGEMVTail
2024-08-14 10:52:46 -05:00
Henry Chen
ef94b96530
Use ldc1 and sdc1 for the prologue and epilogue on LOONGSON3A
...
This fix is similar to
2d8064174c
2024-08-14 18:05:11 +08:00
Martin Kroeker
7ca835a82c
address clang array overflow warning
2024-08-10 13:44:56 +02:00
Martin Kroeker
46e331a917
remove the unworkable GEMM3M restriction from GENERIC again
2024-08-07 19:41:10 +02:00
Martin Kroeker
ccc23338d7
have the dummy GEMM3M kernel at least forward to regular GEMM
2024-08-07 19:39:02 +02:00
Martin Kroeker
f1c9803f9a
add proper return statement
2024-08-04 00:14:31 +02:00
Martin Kroeker
60abcc3991
add proper return statement
2024-08-04 00:13:31 +02:00
Chip Kerchner
1a7b8c650d
Merge branch 'develop' into betterPowerGEMVTail
2024-08-01 14:59:12 -05:00
Martin Kroeker
9afd0c8afd
Merge pull request #4814 from Mousius/gemv-proxy
...
Forward GEMM to GEMV when one argument is actually a vector
2024-07-31 23:18:01 +02:00
Martin Kroeker
edbf093c98
Update zarch SCAL kernels to handle INF and NAN arguments ( #4829 )
...
* handle INF and NAN in input (for S/D only if DUMMY2 argument is set)
2024-07-31 19:45:15 +02:00
Chris Sidebottom
ba2e989c67
Add accumulators to AArch64 GEMV Kernels
...
This helps to reduce values going missing as we accumulate.
2024-07-31 13:09:14 +01:00
Martin Kroeker
a875304eb0
fix inverted conditional for NAN handling
2024-07-26 09:50:20 +02:00
Martin Kroeker
24acdd6bbb
correct offset
2024-07-26 09:49:24 +02:00
Martin Kroeker
fb7c53c5e5
Merge pull request #4807 from martin-frbg/scalfixes
...
[WIP]Make NAN handling in the SCAL kernels depend on the dummy2 parameter
2024-07-25 23:42:50 +02:00
Martin Kroeker
15c53dd2e0
Merge pull request #4794 from XiWeiGu/Fixed_Numpy_CI_Test
...
Try to fixed numpy ci test failures
2024-07-25 23:42:13 +02:00
Martin Kroeker
a4e56e0452
Merge pull request #4806 from Mousius/small-gemm
...
Small GEMM for AArch64 with SVE
2024-07-25 21:50:04 +02:00
yamazaki-mitsufumi
88caf02f62
Fix ambiguous error on Mac OS
2024-07-25 22:43:13 +09:00
Martin Kroeker
b613754143
Update scal..c
2024-07-24 14:31:29 +02:00
Martin Kroeker
f5d04318e3
Merge branch 'OpenMathLib:develop' into scalfixes
2024-07-21 13:43:43 +02:00
Martin Kroeker
73f8866ffb
make NAN handling depend on DUMMY2 parameter
2024-07-21 13:42:47 +02:00
Martin Kroeker
dfbc2348a8
fix NAN handling
2024-07-20 18:27:15 +02:00
