ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-18 03:49:41 +00:00
Code Issues Packages Projects Releases Wiki Activity

MaxPool & AvgPool bwd instances, test, ckProfiler, client example (#861)

Browse Source 
* Add maxpool instances

* Rename index pool to max pool.

* Add maxpool bwd bf16 instances

* Add avg pool bwd instances

* Rename avgpool and maxpool to avg_pool3d and max_pool

* Add bf16 pool fwd instances

* Add max pool bwd to ckProfiler

* Add avg pool3d bwd to ckProfiler

* Add avg pool bwd test

* Fix bug of reference pool fwd (dilation)

* Fix bug of max pool bwd  (dilation and initZero)

* Support bf16 compute data type

* Force compute type be f32. Because atomicAdd only support f32

* Add max pool bwd test

* Rename folder

* Rename pool

* Add max pool bwd client example

* Add avg pool bwd client example

* Add missing workspace

* clang format

* Rename macro

* remove useless header

* remove useless layout

[ROCm/composable_kernel commit: 866377de18]
This commit is contained in:
rocking
2023-08-31 21:01:50 +08:00
committed by GitHub
parent ddf8b9ac4c
commit 0b07461518
45 changed files with 2174 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
example/49_maxpool2d_bwd/maxpool2d_bwd_common.hpp
View File

@@ -8,7 +8,7 @@
#include "ck/ck.hpp"
#include "ck/utility/reduction_enums.hpp"
#include "ck/tensor_operation/gpu/device/impl/device_pool2d_fwd_nhwc_nhwc.hpp"
#include "ck/tensor_operation/gpu/device/impl/device_index_pool_bwd_impl.hpp"
#include "ck/tensor_operation/gpu/device/impl/device_max_pool_bwd_impl.hpp"
#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
#include "ck/library/utility/check_err.hpp"
@@ -60,7 +60,7 @@ bool maxpool_bwd_test(bool do_verification,
1>; // InSrcOutDstVectorSize
using DeviceMaxPoolBwdInstance = ck::tensor_operation::device::
DeviceIndexPoolBwdImpl<DOutDataType, IndexDataType, DInDataType, 4>;
DeviceMaxPoolBwdImpl<DOutDataType, IndexDataType, DInDataType, 4>;
const ck::index_t Ys = (Y - 1) * window_dilation_h + 1;
const ck::index_t Xs = (X - 1) * window_dilation_w + 1;
@@ -155,7 +155,8 @@ bool maxpool_bwd_test(bool do_verification,
dout_n_c_ho_wo.mDesc.GetElementSpaceSize(),
din_n_c_hi_wi_device.mDesc.GetElementSpaceSize(),
window_spatial_lengths,
window_strides);
window_strides,
window_dilations);
if(!pool_bwd.IsSupportedArgument(pool_bwd_argument_ptr.get()))
{