Skip to content

Commit

Permalink
w
Browse files Browse the repository at this point in the history
  • Loading branch information
@nihui
nihui committed Jan 15, 2025
1 parent 137520a commit 1b80186
Show file tree
Hide file tree
Showing 6 changed files with 44 additions and 50 deletions.
42 changes: 14 additions & 28 deletions src/layer/x86/convolution_im2col_gemm_int8.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,29 +13,23 @@
// specific language governing permissions and limitations under the License.

#if NCNN_RUNTIME_CPU && NCNN_AVX512VNNI && __AVX512F__ && !__AVX512VNNI__
// int convolution_im2col_gemm_int8_avx512vnni(const Mat& bottom_blob, Mat& top_blob, const Mat& AT, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h, int nT, const Option& opt);

void convolution_im2col_input_tile_int8_avx512vnni(const Mat& bottom_blob, Mat& B, int j, int max_jj, int k, int max_kk, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h);
#endif

#if NCNN_RUNTIME_CPU && NCNN_AVXVNNI && __AVX__ && !__AVXVNNI__ && !__AVX512VNNI__
// int convolution_im2col_gemm_int8_avxvnni(const Mat& bottom_blob, Mat& top_blob, const Mat& AT, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h, int nT, const Option& opt);
#if NCNN_RUNTIME_CPU && NCNN_AVXVNNIINT8 && __AVX__ && !__AVXVNNIINT8__ && !__AVX512VNNI__
void convolution_im2col_input_tile_int8_avxvnniint8(const Mat& bottom_blob, Mat& B, int j, int max_jj, int k, int max_kk, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h);
#endif

#if NCNN_RUNTIME_CPU && NCNN_AVXVNNI && __AVX__ && !__AVXVNNI__ && !__AVX512VNNI__
void convolution_im2col_input_tile_int8_avxvnni(const Mat& bottom_blob, Mat& B, int j, int max_jj, int k, int max_kk, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h);
#endif

#if NCNN_RUNTIME_CPU && NCNN_AVX2 && __AVX__ && !__AVX2__ && !__AVXVNNI__ && !__AVX512VNNI__
// void convolution_im2col_gemm_transform_kernel_int8_avx2(const Mat& kernel, Mat& AT, int inch, int outch, int kernel_w, int kernel_h, const Option& opt);
// int convolution_im2col_gemm_int8_avx2(const Mat& bottom_blob, Mat& top_blob, const Mat& AT, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h, int nT, const Option& opt);

void convolution_im2col_input_tile_int8_avx2(const Mat& bottom_blob, Mat& B, int j, int max_jj, int k, int max_kk, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h);

void unpack_output_tile_int32_avx2(const Mat& topT, Mat& top_blob, int i, int max_ii, int j, int max_jj);

#endif

#if NCNN_RUNTIME_CPU && NCNN_XOP && __SSE2__ && !__XOP__ && !__AVX2__ && !__AVXVNNI__ && !__AVX512VNNI__
// int convolution_im2col_gemm_int8_xop(const Mat& bottom_blob, Mat& top_blob, const Mat& AT, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h, int nT, const Option& opt);
#endif

// gemm_x86.h
Expand Down Expand Up @@ -2612,13 +2606,13 @@ static void convolution_im2col_input_tile_int8(const Mat& bottom_blob, Mat& B, i
}
#endif

// #if NCNN_RUNTIME_CPU && NCNN_AVXVNNIINT8 && __AVX__ && !__AVXVNNIINT8__ && !__AVX512VNNI__
// if (ncnn::cpu_support_x86_avx_vnni_int8())
// {
// convolution_im2col_input_tile_int8_avxvnniint8(bottom_blob, B, j, max_jj, k, max_kk, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h);
// return;
// }
// #endif
#if NCNN_RUNTIME_CPU && NCNN_AVXVNNIINT8 && __AVX__ && !__AVXVNNIINT8__ && !__AVX512VNNI__
if (ncnn::cpu_support_x86_avx_vnni_int8())
{
convolution_im2col_input_tile_int8_avxvnniint8(bottom_blob, B, j, max_jj, k, max_kk, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h);
return;
}
#endif

#if NCNN_RUNTIME_CPU && NCNN_AVXVNNI && __AVX__ && !__AVXVNNI__ && !__AVXVNNIINT8__ && !__AVX512VNNI__
if (ncnn::cpu_support_x86_avx_vnni())
Expand Down Expand Up @@ -2707,14 +2701,6 @@ static void convolution_im2col_input_tile_int8(const Mat& bottom_blob, Mat& B, i

static void convolution_im2col_gemm_transform_kernel_int8(const Mat& kernel, Mat& AT, int inch, int outch, int kernel_w, int kernel_h, const Option& opt)
{
// #if NCNN_RUNTIME_CPU && NCNN_AVX2 && __AVX__ && !__AVX2__ && !__AVXVNNI__ && !__AVX512VNNI__
// if (ncnn::cpu_support_x86_avx2())
// {
// convolution_im2col_gemm_transform_kernel_int8_avx2(kernel, AT, inch, outch, kernel_w, kernel_h, opt);
// return;
// }
// #endif

// NCNN_LOGE("convolution_im2col_gemm_transform_kernel");
const int maxk = kernel_w * kernel_h;

Expand Down Expand Up @@ -2765,12 +2751,12 @@ static void convolution_im2col_gemm_transform_kernel_int8(const Mat& kernel, Mat
}
}

#if 1 //NCNN_AVX512VNNI || NCNN_AVXVNNI
#if NCNN_AVX512VNNI || NCNN_AVXVNNI
bool has_w_shift = false;
if (TILE_K >= 4)
{
has_w_shift = ncnn::cpu_support_x86_avx512_vnni() || ncnn::cpu_support_x86_avx_vnni();
#if 0 //NCNN_AVXVNNIINT8
#if NCNN_AVXVNNIINT8
if (ncnn::cpu_support_x86_avx_vnni_int8())
has_w_shift = false;
#endif // NCNN_AVXVNNIINT8
Expand Down Expand Up @@ -4585,7 +4571,7 @@ static void unpack_output_tile_int32(const Mat& topT, Mat& top_blob, int i, int
}
}

static int convolution_im2col_gemm_int8(const Mat& bottom_blob, Mat& top_blob, const Mat& AT, const Mat& output_descales, const Mat& bias_data, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h, int nT, const Option& opt)
static int convolution_im2col_gemm_int8(const Mat& bottom_blob, Mat& top_blob, const Mat& AT, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h, int nT, const Option& opt)
{
const int maxk = kernel_w * kernel_h;

Expand Down
3 changes: 1 addition & 2 deletions src/layer/x86/convolution_x86.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1022,8 +1022,7 @@ int Convolution_x86::forward_int8_x86(const Mat& bottom_blob, Mat& top_blob, con
}
else if (opt.use_sgemm_convolution)
{
// ret = convolution_im2col_gemm_int8(bottom_blob_bordered, top_blob_int32, weight_sgemm_data, scale_in_data, bias_data, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h, _nT, opt);
ret = convolution_im2col_gemm_int8(bottom_blob_bordered, top_blob_int32, weight_sgemm_data, scale_in_data, bias_data, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h, _nT, opt);
ret = convolution_im2col_gemm_int8(bottom_blob_bordered, top_blob_int32, weight_sgemm_data, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h, _nT, opt);
}
else
{
Expand Down
10 changes: 0 additions & 10 deletions src/layer/x86/convolution_x86_avx2.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -34,16 +34,6 @@ void convolution_packed_int8_avx2(const Mat& bottom_blob, Mat& top_blob, const M
}

// gemm
// void convolution_im2col_gemm_transform_kernel_int8_avx2(const Mat& kernel, Mat& AT, int inch, int outch, int kernel_w, int kernel_h, const Option& opt)
// {
// convolution_im2col_gemm_transform_kernel_int8(kernel, AT, inch, outch, kernel_w, kernel_h, opt);
// }
//
// int convolution_im2col_gemm_int8_avx2(const Mat& bottom_blob, Mat& top_blob, const Mat& AT, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h, int nT, const Option& opt)
// {
// return convolution_im2col_gemm_int8(bottom_blob, top_blob, AT, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h, nT, opt);
// }

void convolution_im2col_input_tile_int8_avx2(const Mat& bottom_blob, Mat& B, int j, int max_jj, int k, int max_kk, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h)
{
convolution_im2col_input_tile_int8(bottom_blob, B, j, max_jj, k, max_kk, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h);
Expand Down
5 changes: 0 additions & 5 deletions src/layer/x86/convolution_x86_avx512vnni.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -29,11 +29,6 @@ void convolution_packed_int8_avx512vnni(const Mat& bottom_blob, Mat& top_blob, c
}

// gemm
// int convolution_im2col_gemm_int8_avx512vnni(const Mat& bottom_blob, Mat& top_blob, const Mat& AT, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h, int nT, const Option& opt)
// {
// return convolution_im2col_gemm_int8(bottom_blob, top_blob, AT, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h, nT, opt);
// }

void convolution_im2col_input_tile_int8_avx512vnni(const Mat& bottom_blob, Mat& B, int j, int max_jj, int k, int max_kk, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h)
{
convolution_im2col_input_tile_int8(bottom_blob, B, j, max_jj, k, max_kk, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h);
Expand Down
5 changes: 0 additions & 5 deletions src/layer/x86/convolution_x86_avxvnni.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -29,11 +29,6 @@ void convolution_packed_int8_avxvnni(const Mat& bottom_blob, Mat& top_blob, cons
}

// gemm
// int convolution_im2col_gemm_int8_avxvnni(const Mat& bottom_blob, Mat& top_blob, const Mat& AT, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h, int nT, const Option& opt)
// {
// return convolution_im2col_gemm_int8(bottom_blob, top_blob, AT, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h, nT, opt);
// }

void convolution_im2col_input_tile_int8_avxvnni(const Mat& bottom_blob, Mat& B, int j, int max_jj, int k, int max_kk, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h)
{
convolution_im2col_input_tile_int8(bottom_blob, B, j, max_jj, k, max_kk, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h);
Expand Down
29 changes: 29 additions & 0 deletions src/layer/x86/convolution_x86_avxvnniint8.cpp
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,29 @@
// Tencent is pleased to support the open source community by making ncnn available.
//
// Copyright (C) 2024 THL A29 Limited, a Tencent company. All rights reserved.
//
// Licensed under the BSD 3-Clause License (the "License"); you may not use this file except
// in compliance with the License. You may obtain a copy of the License at
//
// https://opensource.org/licenses/BSD-3-Clause
//
// Unless required by applicable law or agreed to in writing, software distributed
// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
// CONDITIONS OF ANY KIND, either express or implied. See the License for the
// specific language governing permissions and limitations under the License.

#include "cpu.h"
#include "mat.h"
#include "x86_usability.h"

namespace ncnn {

#include "convolution_im2col_gemm_int8.h"

// gemm
void convolution_im2col_input_tile_int8_avxvnniint8(const Mat& bottom_blob, Mat& B, int j, int max_jj, int k, int max_kk, int kernel_w, int kernel_h, int dilation_w, int dilation_h, int stride_w, int stride_h)
{
convolution_im2col_input_tile_int8(bottom_blob, B, j, max_jj, k, max_kk, kernel_w, kernel_h, dilation_w, dilation_h, stride_w, stride_h);
}

} // namespace ncnn

0 comments on commit 1b80186

Please sign in to comment.