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
5392333a
Commit
5392333a
authored
Dec 22, 2022
by
rocking
Browse files
Refine var name
parent
441d8973
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
8 additions
and
9 deletions
+8
-9
include/ck/tensor_operation/gpu/grid/gemm_layernorm/gridwise_welford_second_half_layernorm2d.hpp
...mm_layernorm/gridwise_welford_second_half_layernorm2d.hpp
+8
-9
No files found.
include/ck/tensor_operation/gpu/grid/gemm_layernorm/gridwise_welford_second_half_layernorm2d.hpp
View file @
5392333a
...
@@ -69,12 +69,12 @@ struct GridwiseWelfordSecondHalfLayernorm2d
...
@@ -69,12 +69,12 @@ struct GridwiseWelfordSecondHalfLayernorm2d
static
constexpr
auto
thread_buffer_desc_n
=
static
constexpr
auto
thread_buffer_desc_n
=
make_naive_tensor_descriptor_packed
(
make_tuple
(
Number
<
NThreadSliceSize
>
{}));
make_naive_tensor_descriptor_packed
(
make_tuple
(
Number
<
NThreadSliceSize
>
{}));
using
Thread
Reduce
SrcDesc_M_1
=
decltype
(
thread_buffer_desc_m_1
);
using
Thread
Welford
SrcDesc_M_1
=
decltype
(
thread_buffer_desc_m_1
);
using
Thread
Reduce
DstDesc_M
=
using
Thread
Welford
DstDesc_M
=
decltype
(
make_naive_tensor_descriptor_packed
(
make_tuple
(
Number
<
MThreadSliceSize
>
{})));
decltype
(
make_naive_tensor_descriptor_packed
(
make_tuple
(
Number
<
MThreadSliceSize
>
{})));
using
ThreadwiseWelford
=
using
ThreadwiseWelford
=
ThreadwiseWelfordMerge
<
ComputeDataType
,
Thread
Reduce
SrcDesc_M_1
,
Thread
Reduce
DstDesc_M
>
;
ThreadwiseWelfordMerge
<
ComputeDataType
,
Thread
Welford
SrcDesc_M_1
,
Thread
Welford
DstDesc_M
>
;
using
BlockwiseWelford
=
BlockwiseWelford
<
ComputeDataType
,
using
BlockwiseWelford
=
BlockwiseWelford
<
ComputeDataType
,
BlockSize
,
BlockSize
,
...
@@ -298,8 +298,7 @@ struct GridwiseWelfordSecondHalfLayernorm2d
...
@@ -298,8 +298,7 @@ struct GridwiseWelfordSecondHalfLayernorm2d
welford_count_thread_buf
(
I
)
=
0
;
welford_count_thread_buf
(
I
)
=
0
;
});
});
for
(
index_t
reducedTiles
=
0
;
reducedTiles
<
numMeanVarCountBlockTileIteration_N
;
for
(
index_t
n
=
0
;
n
<
numMeanVarCountBlockTileIteration_N
;
++
n
)
++
reducedTiles
)
{
{
threadwise_mean_load_m_nblock
.
Run
(
mean_var_grid_desc_m_n
,
threadwise_mean_load_m_nblock
.
Run
(
mean_var_grid_desc_m_n
,
welford_mean_global_val_buf
,
welford_mean_global_val_buf
,
...
@@ -386,10 +385,10 @@ struct GridwiseWelfordSecondHalfLayernorm2d
...
@@ -386,10 +385,10 @@ struct GridwiseWelfordSecondHalfLayernorm2d
});
});
threadwise_h_store_m_n
.
Run
(
thread_buffer_desc_m_n
,
threadwise_h_store_m_n
.
Run
(
thread_buffer_desc_m_n
,
make_tuple
(
I0
,
I0
),
make_tuple
(
I0
,
I0
),
h_thread_buf
,
h_thread_buf
,
h_grid_desc_m_n
,
h_grid_desc_m_n
,
h_global_val_buf
);
h_global_val_buf
);
}
// run
}
// run
};
};
...
...
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