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
22593950
"...git@developer.sourcefind.cn:cnjsdfcy/simbricks.git" did not exist on "1cb75bf865470bdce9e535a864a07bdf2dfab9a8"
Commit
22593950
authored
Nov 07, 2023
by
letaoqin
Browse files
NXdlPerWave must be one
parent
5a032110
Changes
4
Hide whitespace changes
Inline
Side-by-side
Showing
4 changed files
with
7 additions
and
0 deletions
+7
-0
include/ck/tensor_operation/gpu/grid/gridwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_light_v1.hpp
...dwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_light_v1.hpp
+1
-0
include/ck/tensor_operation/gpu/grid/gridwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_light_v2.hpp
...dwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_light_v2.hpp
+2
-0
include/ck/tensor_operation/gpu/grid/gridwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_v1.hpp
...id/gridwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_v1.hpp
+2
-0
include/ck/tensor_operation/gpu/grid/gridwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_v2.hpp
...id/gridwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_v2.hpp
+2
-0
No files found.
include/ck/tensor_operation/gpu/grid/gridwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_light_v1.hpp
View file @
22593950
...
@@ -1237,6 +1237,7 @@ struct GridwiseBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_Light_V1
...
@@ -1237,6 +1237,7 @@ struct GridwiseBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_Light_V1
static
constexpr
index_t
Size
=
sizeof
(
ck
::
half_t
);
static
constexpr
index_t
Size
=
sizeof
(
ck
::
half_t
);
};
};
static
constexpr
index_t
NThreadClusterLengths
=
32
;
static
constexpr
index_t
NThreadClusterLengths
=
32
;
static_assert
(
NXdlPerWave
==
1
);
static_assert
(
NPerXdl
==
32
);
static_assert
(
NPerXdl
==
32
);
static_assert
(
D0BlockTransferSrcScalarPerVector
*
NThreadClusterLengths
<=
NPerBlock
,
static_assert
(
D0BlockTransferSrcScalarPerVector
*
NThreadClusterLengths
<=
NPerBlock
,
"D0BlockTransferSrcScalarPerVector * NThreadClusterLengths <= NPerBlock"
);
"D0BlockTransferSrcScalarPerVector * NThreadClusterLengths <= NPerBlock"
);
...
...
include/ck/tensor_operation/gpu/grid/gridwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_light_v2.hpp
View file @
22593950
...
@@ -1316,7 +1316,9 @@ struct GridwiseBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_Light_V2
...
@@ -1316,7 +1316,9 @@ struct GridwiseBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_Light_V2
static
constexpr
index_t
Size0
=
0
;
static
constexpr
index_t
Size0
=
0
;
static
constexpr
index_t
Size
=
sizeof
(
ck
::
half_t
);
static
constexpr
index_t
Size
=
sizeof
(
ck
::
half_t
);
};
};
static
constexpr
index_t
NThreadClusterLengths
=
32
;
static
constexpr
index_t
NThreadClusterLengths
=
32
;
static_assert
(
NXdlPerWave
==
1
);
static_assert
(
NPerXdl
==
32
);
static_assert
(
NPerXdl
==
32
);
static_assert
(
D0BlockTransferSrcScalarPerVector
*
NThreadClusterLengths
<=
NPerBlock
,
static_assert
(
D0BlockTransferSrcScalarPerVector
*
NThreadClusterLengths
<=
NPerBlock
,
"D0BlockTransferSrcScalarPerVector * NThreadClusterLengths <= NPerBlock"
);
"D0BlockTransferSrcScalarPerVector * NThreadClusterLengths <= NPerBlock"
);
...
...
include/ck/tensor_operation/gpu/grid/gridwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_v1.hpp
View file @
22593950
...
@@ -1304,7 +1304,9 @@ struct GridwiseBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_V1
...
@@ -1304,7 +1304,9 @@ struct GridwiseBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_V1
static
constexpr
index_t
Size0
=
0
;
static
constexpr
index_t
Size0
=
0
;
static
constexpr
index_t
Size
=
sizeof
(
ck
::
half_t
);
static
constexpr
index_t
Size
=
sizeof
(
ck
::
half_t
);
};
};
static
constexpr
index_t
NThreadClusterLengths
=
32
;
static
constexpr
index_t
NThreadClusterLengths
=
32
;
static_assert
(
NXdlPerWave
==
1
);
static_assert
(
NPerXdl
==
32
);
static_assert
(
NPerXdl
==
32
);
static_assert
(
D0BlockTransferSrcScalarPerVector
*
NThreadClusterLengths
<=
NPerBlock
,
static_assert
(
D0BlockTransferSrcScalarPerVector
*
NThreadClusterLengths
<=
NPerBlock
,
"D0BlockTransferSrcScalarPerVector * NThreadClusterLengths <= NPerBlock"
);
"D0BlockTransferSrcScalarPerVector * NThreadClusterLengths <= NPerBlock"
);
...
...
include/ck/tensor_operation/gpu/grid/gridwise_batched_mha_bwd_xdl_cshuffle_qloop_b2t_v2.hpp
View file @
22593950
...
@@ -1370,7 +1370,9 @@ struct GridwiseBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_V2
...
@@ -1370,7 +1370,9 @@ struct GridwiseBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_V2
static
constexpr
index_t
Size0
=
0
;
static
constexpr
index_t
Size0
=
0
;
static
constexpr
index_t
Size
=
sizeof
(
ck
::
half_t
);
static
constexpr
index_t
Size
=
sizeof
(
ck
::
half_t
);
};
};
static
constexpr
index_t
NThreadClusterLengths
=
32
;
static
constexpr
index_t
NThreadClusterLengths
=
32
;
static_assert
(
NXdlPerWave
==
1
);
static_assert
(
NPerXdl
==
32
);
static_assert
(
NPerXdl
==
32
);
static_assert
(
D0BlockTransferSrcScalarPerVector
*
NThreadClusterLengths
<=
NPerBlock
,
static_assert
(
D0BlockTransferSrcScalarPerVector
*
NThreadClusterLengths
<=
NPerBlock
,
"D0BlockTransferSrcScalarPerVector * NThreadClusterLengths <= NPerBlock"
);
"D0BlockTransferSrcScalarPerVector * NThreadClusterLengths <= NPerBlock"
);
...
...
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