Skip to content

GitLab

Unverified Commit 317d2e2c authored by Illia Silin's avatar Illia Silin Committed by GitHub
Browse files

Merge branch 'develop' into lwpck-1026

parents a245b8f3 1fefd82e
Hide whitespace changes
Inline Side-by-side
Showing with 5 additions and 4 deletions
include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_fwd_multiple_abd_xdl_cshuffle.hpp
View file @ 317d2e2c
...@@ -836,7 +836,7 @@ struct DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle ...@@ -836,7 +836,7 @@ struct DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle
// check if it's 1x1, stride=1 conv // check if it's 1x1, stride=1 conv
for(index_t i = 0; i < NDimSpatial; ++i) for(index_t i = 0; i < NDimSpatial; ++i)
{ {
const index_t X = arg.b_g_k_c_xs_lengths_[i + 2]; const index_t X = arg.b_g_k_c_xs_lengths_[i + 3];
const index_t ConvStride = arg.conv_filter_strides_[i]; const index_t ConvStride = arg.conv_filter_strides_[i];
const index_t LeftPad = arg.input_left_pads_[i]; const index_t LeftPad = arg.input_left_pads_[i];
const index_t RightPad = arg.input_right_pads_[i]; const index_t RightPad = arg.input_right_pads_[i];
...@@ -853,7 +853,7 @@ struct DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle ...@@ -853,7 +853,7 @@ struct DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle
// check if it's 1x1 conv // check if it's 1x1 conv
for(index_t i = 0; i < NDimSpatial; ++i) for(index_t i = 0; i < NDimSpatial; ++i)
{ {
const index_t X = arg.b_g_k_c_xs_lengths_[i + 2]; const index_t X = arg.b_g_k_c_xs_lengths_[i + 3];
const index_t LeftPad = arg.input_left_pads_[i]; const index_t LeftPad = arg.input_left_pads_[i];
const index_t RightPad = arg.input_right_pads_[i]; const index_t RightPad = arg.input_right_pads_[i];
...@@ -1094,6 +1094,7 @@ struct DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle ...@@ -1094,6 +1094,7 @@ struct DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle
<< NXdlPerWave << ", " << NXdlPerWave << ", "
<< ABlockTransferSrcScalarPerVector << ", " << ABlockTransferSrcScalarPerVector << ", "
<< BBlockTransferSrcScalarPerVector << ", " << BBlockTransferSrcScalarPerVector << ", "
<< CDEBlockTransferScalarPerVector_NPerBlock << ", "
<< CShuffleMXdlPerWavePerShuffle << ", " << CShuffleMXdlPerWavePerShuffle << ", "
<< CShuffleNXdlPerWavePerShuffle << CShuffleNXdlPerWavePerShuffle
<< ">"; << ">";
......
include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_fwd_multiple_d_multiple_r_xdl_cshuffle.hpp
View file @ 317d2e2c
...@@ -834,7 +834,7 @@ struct DeviceGroupedConvFwdMultipleDMultipleR_Xdl_CShuffle ...@@ -834,7 +834,7 @@ struct DeviceGroupedConvFwdMultipleDMultipleR_Xdl_CShuffle
// check if it's 1x1, stride=1 conv // check if it's 1x1, stride=1 conv
for(index_t i = 0; i < NDimSpatial; ++i) for(index_t i = 0; i < NDimSpatial; ++i)
{ {
const index_t X = arg.b_g_k_c_xs_lengths_[i + 2]; const index_t X = arg.b_g_k_c_xs_lengths_[i + 3];
const index_t ConvStride = arg.conv_filter_strides_[i]; const index_t ConvStride = arg.conv_filter_strides_[i];
const index_t LeftPad = arg.input_left_pads_[i]; const index_t LeftPad = arg.input_left_pads_[i];
const index_t RightPad = arg.input_right_pads_[i]; const index_t RightPad = arg.input_right_pads_[i];
...@@ -851,7 +851,7 @@ struct DeviceGroupedConvFwdMultipleDMultipleR_Xdl_CShuffle ...@@ -851,7 +851,7 @@ struct DeviceGroupedConvFwdMultipleDMultipleR_Xdl_CShuffle
// check if it's 1x1 conv // check if it's 1x1 conv
for(index_t i = 0; i < NDimSpatial; ++i) for(index_t i = 0; i < NDimSpatial; ++i)
{ {
const index_t X = arg.b_g_k_c_xs_lengths_[i + 2]; const index_t X = arg.b_g_k_c_xs_lengths_[i + 3];
const index_t LeftPad = arg.input_left_pads_[i]; const index_t LeftPad = arg.input_left_pads_[i];
const index_t RightPad = arg.input_right_pads_[i]; const index_t RightPad = arg.input_right_pads_[i];
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment