ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-19 22:39:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

Revise layout of group convolution (#675)

Browse Source 
* [What] Remove pure conv int8 instance
[Why] We will never use pure int8 conv in AI, use int8 quantization instead

* Change layout

* Share the kernel parameter

* Support more type of NHWGC for group conv

* Revise client example of conv 2d, use NHWGC layout

* Add instance to cmake

* Revise layout of group conv quantization instance

* Revise layout of external api of group conv quantization

* Revise layout of group conv quantization client example

* Fix clang format

* Add comment to describe meaning of each parameter
This commit is contained in:
rocking
2023-04-24 12:40:00 +08:00
committed by GitHub
parent 903cd19ce3
commit 3eecbfb6ec
41 changed files with 1079 additions and 1222 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
example/40_conv2d_fwd_quantization/run_conv2d_fwd_bias_perchannel_quantization_example.inc
View File

@@ -190,11 +190,11 @@ int run_conv2d_fwd_bias_perchannel_quantization_example(const OutElementOp& out_
const auto in_element_op = InElementOp{};
const auto wei_element_op = WeiElementOp{};
using InLayout = ck::tensor_layout::convolution::GNHWC;
using WeiLayout = ck::tensor_layout::convolution::GKYXC;
using InLayout = ck::tensor_layout::convolution::NHWGC;
using WeiLayout = ck::tensor_layout::convolution::KYXGC;
using BiasLayout = ck::tensor_layout::convolution::G_K;
using RequantScaleLayout = ck::tensor_layout::convolution::G_K;
using OutLayout = ck::tensor_layout::convolution::GNHWK;
using OutLayout = ck::tensor_layout::convolution::NHWGK;
const auto in_g_n_c_wis_desc =
ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(conv_param);

6
example/40_conv2d_fwd_quantization/run_conv2d_fwd_bias_perlayer_quantization_example.inc
View File

@@ -178,10 +178,10 @@ int run_conv2d_fwd_bias_perlayer_quantization_example(const OutElementOp& out_el
const auto in_element_op = InElementOp{};
const auto wei_element_op = WeiElementOp{};
using InLayout = ck::tensor_layout::convolution::GNHWC;
using WeiLayout = ck::tensor_layout::convolution::GKYXC;
using InLayout = ck::tensor_layout::convolution::NHWGC;
using WeiLayout = ck::tensor_layout::convolution::KYXGC;
using BiasLayout = ck::tensor_layout::convolution::G_K;
using OutLayout = ck::tensor_layout::convolution::GNHWK;
using OutLayout = ck::tensor_layout::convolution::NHWGK;
const auto in_g_n_c_wis_desc =
ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(conv_param);

6
example/40_conv2d_fwd_quantization/run_conv2d_fwd_perchannel_quantization_example.inc
View File

@@ -180,10 +180,10 @@ int run_conv2d_fwd_perchannel_quantization_example(const OutElementOp& out_eleme
const auto in_element_op = InElementOp{};
const auto wei_element_op = WeiElementOp{};
using InLayout = ck::tensor_layout::convolution::GNHWC;
using WeiLayout = ck::tensor_layout::convolution::GKYXC;
using InLayout = ck::tensor_layout::convolution::NHWGC;
using WeiLayout = ck::tensor_layout::convolution::KYXGC;
using RequantScaleLayout = ck::tensor_layout::convolution::G_K;
using OutLayout = ck::tensor_layout::convolution::GNHWK;
using OutLayout = ck::tensor_layout::convolution::NHWGK;
const auto in_g_n_c_wis_desc =
ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(conv_param);

6
example/40_conv2d_fwd_quantization/run_conv2d_fwd_perlayer_quantization_example.inc
View File

@@ -162,9 +162,9 @@ int run_conv2d_fwd_perlayer_quantization_example(const OutElementOp& out_element
const auto in_element_op = InElementOp{};
const auto wei_element_op = WeiElementOp{};
using InLayout = ck::tensor_layout::convolution::GNHWC;
using WeiLayout = ck::tensor_layout::convolution::GKYXC;
using OutLayout = ck::tensor_layout::convolution::GNHWK;
using InLayout = ck::tensor_layout::convolution::NHWGC;
using WeiLayout = ck::tensor_layout::convolution::KYXGC;
using OutLayout = ck::tensor_layout::convolution::NHWGK;
const auto in_g_n_c_wis_desc =
ck::utils::conv::make_input_host_tensor_descriptor_g_n_c_wis_packed<InLayout>(conv_param);