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
2ebc3248
Commit
2ebc3248
authored
Jan 13, 2023
by
fsx950223
Browse files
change global function name
parent
09a717e3
Changes
2
Hide whitespace changes
Inline
Side-by-side
Showing
2 changed files
with
11 additions
and
11 deletions
+11
-11
include/ck/tensor_operation/gpu/device/impl/device_batched_gemm_softmax_gemm_xdl_cshuffle.hpp
...ce/impl/device_batched_gemm_softmax_gemm_xdl_cshuffle.hpp
+2
-2
include/ck/tensor_operation/gpu/device/impl/device_grouped_multihead_attention_backward_xdl_cshuffle.hpp
...ice_grouped_multihead_attention_backward_xdl_cshuffle.hpp
+9
-9
No files found.
include/ck/tensor_operation/gpu/device/impl/device_batched_gemm_softmax_gemm_xdl_cshuffle.hpp
View file @
2ebc3248
...
...
@@ -42,7 +42,7 @@ __global__ void
#if CK_USE_LAUNCH_BOUNDS
__launch_bounds__
(
CK_MAX_THREAD_PER_BLOCK
,
CK_MIN_BLOCK_PER_CU
)
#endif
kernel_batched_
gemm_softmax_gemm
_xdl_cshuffle_v1
(
kernel_batched_
multihead_attention_backward
_xdl_cshuffle_v1
(
const
FloatAB
*
__restrict__
p_a_grid
,
const
FloatAB
*
__restrict__
p_b_grid
,
const
FloatAB
*
__restrict__
p_b1_grid
,
...
...
@@ -540,7 +540,7 @@ struct DeviceBatchedGemmSoftmaxGemm_Xdl_CShuffle
float
ave_time
=
0
;
auto
launch_kernel
=
[
&
](
auto
has_main_k_block_loop_
)
{
const
auto
kernel
=
kernel_batched_
gemm_softmax_gemm
_xdl_cshuffle_v1
<
const
auto
kernel
=
kernel_batched_
multihead_attention_backward
_xdl_cshuffle_v1
<
GridwiseGemm
,
ADataType
,
// TODO: distiguish A/B datatype
CDataType
,
...
...
include/ck/tensor_operation/gpu/device/impl/device_grouped_multihead_attention_backward_xdl_cshuffle.hpp
View file @
2ebc3248
...
...
@@ -38,7 +38,7 @@ __global__ void
#if CK_USE_LAUNCH_BOUNDS
__launch_bounds__
(
CK_MAX_THREAD_PER_BLOCK
,
CK_MIN_BLOCK_PER_CU
)
#endif
kernel_grouped_
gemm_softmax_gemm
_xdl_cshuffle_v1
(
kernel_grouped_
multihead_attention_backward
_xdl_cshuffle_v1
(
const
void
CK_CONSTANT_ADDRESS_SPACE
*
group_kernel_args
,
const
index_t
group_count
,
const
AElementwiseOperation
a_element_op
,
...
...
@@ -890,14 +890,14 @@ struct DeviceGroupedMultiheadAttentionBackward_Xdl_CShuffle
auto
launch_kernel
=
[
&
](
auto
has_main_k_block_loop_
)
{
const
auto
kernel
=
kernel_grouped_
gemm_softmax_gemm
_xdl_cshuffle_v1
<
GridwiseGemm
,
GroupKernelArg
,
AElementwiseOperation
,
BElementwiseOperation
,
AccElementwiseOperation
,
B1ElementwiseOperation
,
CElementwiseOperation
,
has_main_k_block_loop_
>
;
kernel_grouped_
multihead_attention_backward
_xdl_cshuffle_v1
<
GridwiseGemm
,
GroupKernelArg
,
AElementwiseOperation
,
BElementwiseOperation
,
AccElementwiseOperation
,
B1ElementwiseOperation
,
CElementwiseOperation
,
has_main_k_block_loop_
>
;
return
launch_and_time_kernel
(
stream_config
,
...
...
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