Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[QNNPACK, Sparsity] Add wrapper for the kernels #51150

Closed
wants to merge 4 commits into from
Closed

[QNNPACK, Sparsity] Add wrapper for the kernels #51150

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
23a97d6
[QNNPACK, Sparsity] Add wrapper for the kernels
kimishpatel Jan 26, 2021
f966ce4
Update on "[QNNPACK, Sparsity] Add wrapper for the kernels"
kimishpatel Jan 26, 2021
78d8d45
Update on "[QNNPACK, Sparsity] Add wrapper for the kernels"
kimishpatel Jan 27, 2021
7cf449b
Update on "[QNNPACK, Sparsity] Add wrapper for the kernels"
kimishpatel Jan 29, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
4 changes: 1 addition & 3 deletions aten/src/ATen/native/quantized/cpu/qnnpack/src/q8gemm_sparse/8x4-packA-sse2.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,9 +11,7 @@
#include <qnnpack/q8gemm_sparse.h>
#include <requantization/runtime-sse2.h>

#define MR 8
#define COL_BLOCK_SIZE 4
#define PACKED_A_BLOCK_SIZE COL_BLOCK_SIZE*MR
#include "8x4c1x4-packed-sse2.h"

// This is a super slow kernel in that it does not use intrinsics to
// tranpose. Since this is for x86 we are not optimizing it.
Expand Down
4 changes: 1 addition & 3 deletions aten/src/ATen/native/quantized/cpu/qnnpack/src/q8gemm_sparse/8x4c1x4-dq-packedA-sse2.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,9 +11,7 @@
#include <qnnpack/q8gemm_sparse.h>
#include <requantization/runtime-sse2.h>

#define MR 8
#define COL_BLOCK_SIZE 4
#define PACKED_A_BLOCK_SIZE COL_BLOCK_SIZE*MR
#include "8x4c1x4-packed-sse2.h"

#define CONVERT_TO_FP_AND_TRANSPOSE(a, b, c, d, t_a, t_b, t_c, t_d) \
a_ps = _mm_cvtepi32_ps(a); \
Expand Down
5 changes: 5 additions & 0 deletions aten/src/ATen/native/quantized/cpu/qnnpack/src/q8gemm_sparse/8x4c1x4-packed-sse2.h
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,5 @@
#pragma once

#define MR 8
#define COL_BLOCK_SIZE 4
#define PACKED_A_BLOCK_SIZE COL_BLOCK_SIZE*MR
3 changes: 3 additions & 0 deletions aten/src/ATen/native/quantized/cpu/qnnpack/wrappers/q8gemm_sparse/4x4-packA-aarch32-neon.S
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,3 @@
#if defined(__arm__)
#include <q8gemm_sparse/4x4-packA-aarch32-neon.S>
#endif /* defined(__arm__) */
3 changes: 3 additions & 0 deletions ...Ten/native/quantized/cpu/qnnpack/wrappers/q8gemm_sparse/4x8c1x4-dq-packedA-aarch32-neon.S
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,3 @@
#if defined(__arm__)
#include <q8gemm_sparse/4x8c1x4-dq-packedA-aarch32-neon.S>
#endif /* defined(__arm__) */
3 changes: 3 additions & 0 deletions ...Ten/native/quantized/cpu/qnnpack/wrappers/q8gemm_sparse/4x8c8x1-dq-packedA-aarch32-neon.S
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,3 @@
#if defined(__arm__)
#include <q8gemm_sparse/4x8c8x1-dq-packedA-aarch32-neon.S>
#endif /* defined(__arm__) */
3 changes: 3 additions & 0 deletions aten/src/ATen/native/quantized/cpu/qnnpack/wrappers/q8gemm_sparse/8x4-packA-aarch64-neon.S
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,3 @@
#if defined(__aarch64__)
#include <q8gemm_sparse/8x4-packA-aarch64-neon.S>
#endif /* defined(__arm__) */
4 changes: 4 additions & 0 deletions aten/src/ATen/native/quantized/cpu/qnnpack/wrappers/q8gemm_sparse/8x4c1x4-packed-sse2.c
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,4 @@
#if defined(__i386__) || defined(__i686__) || defined(__x86_64__)
#include <q8gemm_sparse/8x4-packA-sse2.c>
#include <q8gemm_sparse/8x4c1x4-dq-packedA-sse2.c>
#endif /* defined(__i386__) || defined(__i686__) || defined(__x86_64__) */
3 changes: 3 additions & 0 deletions ...Ten/native/quantized/cpu/qnnpack/wrappers/q8gemm_sparse/8x8c1x4-dq-packedA-aarch64-neon.S
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,3 @@
#if defined(__aarch64__)
#include <q8gemm_sparse/8x8c1x4-dq-packedA-aarch64-neon.S>
#endif /* defined(__arm__) */
3 changes: 3 additions & 0 deletions ...Ten/native/quantized/cpu/qnnpack/wrappers/q8gemm_sparse/8x8c8x1-dq-packedA-aarch64-neon.S
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,3 @@
#if defined(__aarch64__)
#include <q8gemm_sparse/8x8c8x1-dq-packedA-aarch64-neon.S>
#endif /* defined(__arm__) */