frozenleaves/pytorch
1
0
Fork 0
mirror of https://github.com/pytorch/pytorch.git synced 2025-10-20 12:54:11 +08:00
Code Issues Packages Projects Releases Wiki Activity

[PyTorch][XNNPACK] Update wrappers for internal only x86 SSE2 kernels (#96896)

Browse Source 
Summary:
Same as D43747173 (https://github.com/pytorch/pytorch/pull/95911) except for the newly added x86 SSE2 kernels.

For future reference, wrappers can be generated by

```
cd ~/fbsource/xplat/third-party/XNNPACK
# Update the list of internal only kernels in generate-wrappers.py
python3 generate-wrappers.py
```

Test Plan: CI

Reviewed By: digantdesai

Differential Revision: D44072764

Pull Request resolved: https://github.com/pytorch/pytorch/pull/96896
Approved by: https://github.com/digantdesai
This commit is contained in:
Stephen Jia
2023-03-17 03:07:39 +00:00
committed by PyTorch MergeBot
parent f865e23abc
commit 4de1bc16e3
2 changed files with 18 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
third_party/xnnpack_src_defs.bzl vendored
View File

@ -348,8 +348,8 @@ AARCH64_ASM_MICROKERNEL_SRCS = [
"XNNPACK/src/qu8-igemm/gen/qu8-igemm-4x16c4-minmax-rndnu-asm-aarch64-neondot-cortex-a55.S",
"XNNPACK/src/qu8-igemm/gen/qu8-igemm-4x16c4-minmax-rndnu-asm-aarch64-neondot-ld128.S",
] + [
"xnnpack_wrappers/qc8-gemm/gen/qc8-gemm-4x16c4-minmax-none-asm-aarch64-neondot-cortex-a55.S",
"xnnpack_wrappers/qs8-gemm/gen/qs8-gemm-4x16c4-minmax-none-asm-aarch64-neondot-cortex-a55.S",
"XNNPACK/src/qc8-gemm/gen/qc8-gemm-4x16c4-minmax-none-asm-aarch64-neondot-cortex-a55.S",
"XNNPACK/src/qs8-gemm/gen/qs8-gemm-4x16c4-minmax-none-asm-aarch64-neondot-cortex-a55.S",
] if native.read_config("pt", "is_oss", "0") == "0" else []
ALL_ARMSIMD32_MICROKERNEL_SRCS = [
@ -5094,7 +5094,14 @@ ALL_SSE2_MICROKERNEL_SRCS = [
"XNNPACK/src/x64-transposec/gen/x64-transposec-2x2-reuse-switch-sse2.c",
"XNNPACK/src/xx-fill/xx-fill-sse2-x64.c",
"XNNPACK/src/xx-pad/xx-pad-sse2.c",
]
] + [
"XNNPACK/src/qc8-gemm/gen/qc8-gemm-1x4c8-minmax-none-sse2-ld64.c",
"XNNPACK/src/qc8-gemm/gen/qc8-gemm-2x4c8-minmax-none-sse2-ld64.c",
"XNNPACK/src/qc8-gemm/gen/qc8-gemm-3x4c8-minmax-none-sse2-ld64.c",
"XNNPACK/src/qs8-gemm/gen/qs8-gemm-1x4c8-minmax-none-sse2-ld64.c",
"XNNPACK/src/qs8-gemm/gen/qs8-gemm-2x4c8-minmax-none-sse2-ld64.c",
"XNNPACK/src/qs8-gemm/gen/qs8-gemm-3x4c8-minmax-none-sse2-ld64.c",
] if native.read_config("pt", "is_oss", "0") == "0" else []
ALL_SSE41_MICROKERNEL_SRCS = [
"XNNPACK/src/f16-f32-vcvt/gen/f16-f32-vcvt-sse41-int16-x8.c",

9
third_party/xnnpack_wrapper_defs.bzl vendored
View File

@ -5082,7 +5082,14 @@ ALL_SSE2_MICROKERNEL_SRCS = [
"xnnpack_wrappers/x64-transposec/gen/x64-transposec-2x2-reuse-switch-sse2.c",
"xnnpack_wrappers/xx-fill/xx-fill-sse2-x64.c",
"xnnpack_wrappers/xx-pad/xx-pad-sse2.c",
]
] + [
"xnnpack_wrappers/qc8-gemm/gen/qc8-gemm-1x4c8-minmax-none-sse2-ld64.c",
"xnnpack_wrappers/qc8-gemm/gen/qc8-gemm-2x4c8-minmax-none-sse2-ld64.c",
"xnnpack_wrappers/qc8-gemm/gen/qc8-gemm-3x4c8-minmax-none-sse2-ld64.c",
"xnnpack_wrappers/qs8-gemm/gen/qs8-gemm-1x4c8-minmax-none-sse2-ld64.c",
"xnnpack_wrappers/qs8-gemm/gen/qs8-gemm-2x4c8-minmax-none-sse2-ld64.c",
"xnnpack_wrappers/qs8-gemm/gen/qs8-gemm-3x4c8-minmax-none-sse2-ld64.c",
] if native.read_config("pt", "is_oss", "0") == "0" else []
ALL_SSE41_MICROKERNEL_SRCS = [
"xnnpack_wrappers/f16-f32-vcvt/gen/f16-f32-vcvt-sse41-int16-x8.c",