Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
gaoqiong
composable_kernel
Commits
5cfd01fd
"runner/vscode:/vscode.git/clone" did not exist on "f95a1f2bef74cc5badd12b8a6f873d47298595ad"
Commit
5cfd01fd
authored
Aug 20, 2021
by
ltqin
Browse files
format
parent
1e9c511c
Changes
2
Show whitespace changes
Inline
Side-by-side
Showing
2 changed files
with
34 additions
and
40 deletions
+34
-40
host/driver_offline/include/device_convolution_backward_weight_implicit_gemm_v4r4r2_xdlops_nchw_kcyx_nkhw.hpp
...ard_weight_implicit_gemm_v4r4r2_xdlops_nchw_kcyx_nkhw.hpp
+11
-15
host/driver_offline/src/conv_wrw_driver_offline.cpp
host/driver_offline/src/conv_wrw_driver_offline.cpp
+23
-25
No files found.
host/driver_offline/include/device_convolution_backward_weight_implicit_gemm_v4r4r2_xdlops_nchw_kcyx_nkhw.hpp
View file @
5cfd01fd
...
@@ -77,8 +77,8 @@ void device_convolution_backward_weight_implicit_gemm_v4r4r2_xdlops_nchw_kcyx_nk
...
@@ -77,8 +77,8 @@ void device_convolution_backward_weight_implicit_gemm_v4r4r2_xdlops_nchw_kcyx_nk
constexpr
index_t
GemmCThreadTransferDstScalarPerVector
=
1
;
constexpr
index_t
GemmCThreadTransferDstScalarPerVector
=
1
;
#endif
#endif
const
auto
descs
=
const
auto
descs
=
transform_backward_weight_convolution_into_gemm_v4r4r2_nchw_kcyx_nkhw_pad
(
transform_backward_weight_convolution_into_gemm_v4r4r2_nchw_kcyx_nkhw_pad
(
wei_k_c_y_x_desc
,
wei_k_c_y_x_desc
,
in_n_c_hi_wi_desc
,
in_n_c_hi_wi_desc
,
out_n_k_ho_wo_desc
,
out_n_k_ho_wo_desc
,
conv_strides
,
conv_strides
,
...
@@ -93,13 +93,9 @@ void device_convolution_backward_weight_implicit_gemm_v4r4r2_xdlops_nchw_kcyx_nk
...
@@ -93,13 +93,9 @@ void device_convolution_backward_weight_implicit_gemm_v4r4r2_xdlops_nchw_kcyx_nk
// HACK: hacks that control index calculation when iterating over A, B, C matrix
// HACK: hacks that control index calculation when iterating over A, B, C matrix
constexpr
auto
out_gemmk0_gemmm_gemmk1_grid_step_hacks
=
make_tuple
(
constexpr
auto
out_gemmk0_gemmm_gemmk1_grid_step_hacks
=
make_tuple
(
make_tuple
(
Sequence
<
0
,
0
,
1
,
0
,
0
>
{},
make_tuple
(
Sequence
<
0
,
0
,
1
,
0
,
0
>
{},
Sequence
<
0
,
0
,
0
,
0
,
0
>
{},
Sequence
<
0
,
0
,
1
,
0
,
0
>
{}),
Sequence
<
0
,
0
,
0
,
0
,
0
>
{},
Sequence
<
0
,
0
,
1
,
0
,
0
>
{}),
make_tuple
(
make_tuple
(
Sequence
<
0
,
0
,
2
,
0
,
0
>
{},
Sequence
<
0
,
0
,
2
,
0
,
0
>
{},
Sequence
<
0
,
0
,
0
,
0
,
0
>
{},
Sequence
<
0
,
0
,
2
,
0
,
0
>
{}));
Sequence
<
0
,
0
,
0
,
0
,
0
>
{},
Sequence
<
0
,
0
,
2
,
0
,
0
>
{}));
constexpr
auto
in_gemmk0_gemmn_gemmk1_grid_step_hacks
=
constexpr
auto
in_gemmk0_gemmn_gemmk1_grid_step_hacks
=
make_tuple
(
make_tuple
(
Sequence
<
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
1
,
0
,
0
>
{},
make_tuple
(
make_tuple
(
Sequence
<
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
1
,
0
,
0
>
{},
...
...
host/driver_offline/src/conv_wrw_driver_offline.cpp
View file @
5cfd01fd
...
@@ -14,7 +14,6 @@
...
@@ -14,7 +14,6 @@
#include "device_tensor.hpp"
#include "device_tensor.hpp"
#include "device_convolution_backward_weight_implicit_gemm_v4r4r2_xdlops_nchw_kcyx_nkhw.hpp"
#include "device_convolution_backward_weight_implicit_gemm_v4r4r2_xdlops_nchw_kcyx_nkhw.hpp"
#define USE_MODE 1
#define USE_MODE 1
#define USE_CONV_WRW_V4R4R2_XDL_NCHW 1
#define USE_CONV_WRW_V4R4R2_XDL_NCHW 1
...
@@ -245,7 +244,6 @@ int main(int argc, char* argv[])
...
@@ -245,7 +244,6 @@ int main(int argc, char* argv[])
in_right_pads_dev
);
in_right_pads_dev
);
};
};
#if USE_CONV_WRW_V4R4R2_XDL_NCHW
#if USE_CONV_WRW_V4R4R2_XDL_NCHW
if
(
algo
==
ConvBackwardWeightAlgo
::
V4R4R2XDLNCHW
)
if
(
algo
==
ConvBackwardWeightAlgo
::
V4R4R2XDLNCHW
)
{
{
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment