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
dad4cfe3
Commit
dad4cfe3
authored
May 10, 2023
by
Jing Zhang
Browse files
added more instances into dl_gemm_example
parent
b076a02a
Changes
4
Hide whitespace changes
Inline
Side-by-side
Showing
4 changed files
with
17 additions
and
9 deletions
+17
-9
example/01_gemm/common.hpp
example/01_gemm/common.hpp
+6
-6
example/01_gemm/gemm_dl_fp16.cpp
example/01_gemm/gemm_dl_fp16.cpp
+9
-1
include/ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp
...de/ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp
+1
-1
script/cmake-ck-dev.sh
script/cmake-ck-dev.sh
+1
-1
No files found.
example/01_gemm/common.hpp
View file @
dad4cfe3
...
@@ -24,13 +24,13 @@
...
@@ -24,13 +24,13 @@
struct
ProblemSize
final
struct
ProblemSize
final
{
{
ck
::
index_t
M
=
3840
;
ck
::
index_t
M
=
1024
;
ck
::
index_t
N
=
4096
;
ck
::
index_t
N
=
1024
;
ck
::
index_t
K
=
409
6
;
ck
::
index_t
K
=
25
6
;
ck
::
index_t
StrideA
=
4096
;
ck
::
index_t
StrideA
=
M
;
ck
::
index_t
StrideB
=
4096
;
ck
::
index_t
StrideB
=
N
;
ck
::
index_t
StrideC
=
4096
;
ck
::
index_t
StrideC
=
N
;
};
};
struct
ExecutionConfig
final
struct
ExecutionConfig
final
...
...
example/01_gemm/gemm_dl_fp16.cpp
View file @
dad4cfe3
...
@@ -26,7 +26,15 @@ using DeviceGemmInstance = ck::tensor_operation::device::DeviceGemmDl
...
@@ -26,7 +26,15 @@ using DeviceGemmInstance = ck::tensor_operation::device::DeviceGemmDl
// ######| Type| Type| Type| Type| | | | Elementwise| Elementwise| Elementwise| Spacialization| Size| Block| Block| Block| | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths| ThreadCluster| SrcAccess| SrcVectorTensor| SrcVectorTensor| DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths| ThreadCluster| SrcAccess| SrcVectorTensor| SrcVectorTensor| DstVectorTensor| SrcDstAccess| SrcDstVectorDim| DstScalarPerVector|
// ######| Type| Type| Type| Type| | | | Elementwise| Elementwise| Elementwise| Spacialization| Size| Block| Block| Block| | ThreadM111| ThreadN111| Thread| ClusterM110Xs| ClusterN110Xs| ThreadSliceLengths| ThreadClusterLengths| ThreadCluster| SrcAccess| SrcVectorTensor| SrcVectorTensor| DstVectorTensor| ThreadSliceLengths| ThreadClusterLengths| ThreadCluster| SrcAccess| SrcVectorTensor| SrcVectorTensor| DstVectorTensor| SrcDstAccess| SrcDstVectorDim| DstScalarPerVector|
// ######| | | | | | | | Operation| Operation| Operation| | | | | | | | | | | | K0_M0_M1_K1| K0_M0_M1_K1| ArrangeOrder| Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder| Lengths_K0_M0_M1_K1| K0_N0_N1_K1| K0_N0_N1_K1| ArrangeOrder| Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder| Lengths_K0_N0_N1_K1| Order| | |
// ######| | | | | | | | Operation| Operation| Operation| | | | | | | | | | | | K0_M0_M1_K1| K0_M0_M1_K1| ArrangeOrder| Order| Lengths_K0_M0_M1_K1| ContiguousDimOrder| Lengths_K0_M0_M1_K1| K0_N0_N1_K1| K0_N0_N1_K1| ArrangeOrder| Order| Lengths_K0_N0_N1_K1| ContiguousDimOrder| Lengths_K0_N0_N1_K1| Order| | |
// ######| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
// ######| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
<
ADataType
,
BDataType
,
CDataType
,
AccDataType
,
ALayout
,
BLayout
,
CLayout
,
AElementOp
,
BElementOp
,
CElementOp
,
GemmDefault
,
256
,
128
,
128
,
16
,
2
,
4
,
4
,
1
,
S
<
8
,
2
>
,
S
<
8
,
2
>
,
S
<
2
,
1
,
4
,
2
>
,
S
<
8
,
1
,
32
,
1
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
1
,
1
,
4
,
1
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
1
,
1
,
4
,
2
>
,
S
<
2
,
1
,
4
,
2
>
,
S
<
8
,
1
,
32
,
1
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
1
,
1
,
4
,
1
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
1
,
1
,
4
,
2
>
,
S
<
0
,
1
,
2
,
3
,
4
,
5
>
,
5
,
4
>
;
<
ADataType
,
BDataType
,
CDataType
,
AccDataType
,
ALayout
,
BLayout
,
CLayout
,
AElementOp
,
BElementOp
,
CElementOp
,
GemmDefault
,
256
,
128
,
128
,
8
,
2
,
4
,
4
,
1
,
S
<
8
,
2
>
,
S
<
8
,
2
>
,
S
<
1
,
1
,
4
,
2
>
,
S
<
8
,
1
,
32
,
1
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
1
,
1
,
4
,
1
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
1
,
1
,
4
,
2
>
,
S
<
1
,
1
,
4
,
2
>
,
S
<
8
,
1
,
32
,
1
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
1
,
1
,
4
,
1
>
,
S
<
0
,
3
,
1
,
2
>
,
S
<
1
,
1
,
4
,
2
>
,
S
<
0
,
1
,
2
,
3
,
4
,
5
>
,
5
,
4
>
;
//< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmDefault, 128, 64, 128, 8, 2, 4, 4, 1, S<4, 2>, S<8, 2>, S<1, 1, 4, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 4, 1>, S<0, 3, 1, 2>, S<1, 1, 4, 2>, S<1, 1, 8, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 4, 1>, S<0, 3, 1, 2>, S<1, 1, 4, 2>, S<0, 1, 2, 3, 4, 5>, 5, 4>;
//< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmDefault, 128, 128, 64, 8, 2, 4, 4, 1, S<8, 2>, S<4, 2>, S<1, 1, 8, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 4, 1>, S<0, 3, 1, 2>, S<1, 1, 4, 2>, S<1, 1, 4, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 4, 1>, S<0, 3, 1, 2>, S<1, 1, 4, 2>, S<0, 1, 2, 3, 4, 5>, 5, 4>;
//< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmDefault, 256, 64, 128, 8, 2, 2, 4, 1, S<8, 2>, S<8, 2>, S<1, 1, 2, 2>, S<8, 1, 32, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 2, 1>, S<0, 3, 1, 2>, S<1, 1, 2, 2>, S<1, 1, 4, 2>, S<8, 1, 32, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 4, 1>, S<0, 3, 1, 2>, S<1, 1, 4, 2>, S<0, 1, 2, 3, 4, 5>, 5, 4>;
//< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmDefault, 256, 128, 64, 8, 2, 4, 2, 1, S<8, 2>, S<8, 2>, S<1, 1, 4, 2>, S<8, 1, 32, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 4, 1>, S<0, 3, 1, 2>, S<1, 1, 4, 2>, S<1, 1, 2, 2>, S<8, 1, 32, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 2, 1>, S<0, 3, 1, 2>, S<1, 1, 2, 2>, S<0, 1, 2, 3, 4, 5>, 5, 2>;
//< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmDefault, 128, 32, 128, 8, 2, 2, 4, 1, S<4, 2>, S<8, 2>, S<1, 1, 2, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 2, 1>, S<0, 3, 1, 2>, S<1, 1, 2, 2>, S<1, 1, 8, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 4, 1>, S<0, 3, 1, 2>, S<1, 1, 4, 2>, S<0, 1, 2, 3, 4, 5>, 5, 4>;
//< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmDefault, 128, 128, 32, 8, 2, 4, 2, 1, S<8, 2>, S<4, 2>, S<1, 1, 8, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 4, 1>, S<0, 3, 1, 2>, S<1, 1, 4, 2>, S<1, 1, 2, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 2, 1>, S<0, 3, 1, 2>, S<1, 1, 2, 2>, S<0, 1, 2, 3, 4, 5>, 5, 2>;
//< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmDefault, 128, 32, 64, 8, 2, 2, 2, 1, S<4, 2>, S<8, 2>, S<1, 1, 2, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 2, 1>, S<0, 3, 1, 2>, S<1, 1, 2, 2>, S<1, 1, 4, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 4, 1>, S<0, 3, 1, 2>, S<1, 1, 4, 2>, S<0, 1, 2, 3, 4, 5>, 5, 4>;
//< ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout, AElementOp, BElementOp, CElementOp, GemmDefault, 128, 64, 32, 8, 2, 2, 2, 1, S<8, 2>, S<4, 2>, S<1, 1, 4, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 4, 1>, S<0, 3, 1, 2>, S<1, 1, 4, 2>, S<1, 1, 2, 2>, S<8, 1, 16, 1>, S<0, 3, 1, 2>, S<0, 3, 1, 2>, S<1, 1, 2, 1>, S<0, 3, 1, 2>, S<1, 1, 2, 2>, S<0, 1, 2, 3, 4, 5>, 5, 2>;
// clang-format on
// clang-format on
using
ReferenceGemmInstance
=
ck
::
tensor_operation
::
host
::
using
ReferenceGemmInstance
=
ck
::
tensor_operation
::
host
::
...
...
include/ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp
View file @
dad4cfe3
...
@@ -485,7 +485,7 @@ struct DeviceGemmDl : public DeviceGemm<ALayout,
...
@@ -485,7 +485,7 @@ struct DeviceGemmDl : public DeviceGemm<ALayout,
static
bool
IsSupportedArgument
(
const
Argument
&
arg
)
static
bool
IsSupportedArgument
(
const
Argument
&
arg
)
{
{
if
(
ck
::
get_device_name
()
==
"gfx906"
||
ck
::
get_device_name
()
==
"gfx1030"
)
if
(
ck
::
get_device_name
()
==
"gfx906"
||
ck
::
get_device_name
()
==
"gfx1030"
||
ck
::
get_device_name
()
==
"gfx1100"
)
{
{
return
GridwiseGemm
::
CheckValidity
(
return
GridwiseGemm
::
CheckValidity
(
arg
.
a_grid_desc_k0_m_k1_
,
arg
.
b_grid_desc_k0_n_k1_
,
arg
.
c_grid_desc_m_n_
);
arg
.
a_grid_desc_k0_m_k1_
,
arg
.
b_grid_desc_k0_n_k1_
,
arg
.
c_grid_desc_m_n_
);
...
...
script/cmake-ck-dev.sh
View file @
dad4cfe3
...
@@ -12,7 +12,7 @@ cmake
...
@@ -12,7 +12,7 @@ cmake
-save-temps=
$PWD
"
\
-save-temps=
$PWD
"
\
-D
CMAKE_BUILD_TYPE
=
Release
\
-D
CMAKE_BUILD_TYPE
=
Release
\
-D
BUILD_DEV
=
ON
\
-D
BUILD_DEV
=
ON
\
-D
GPU_TARGETS
=
"gfx
908;gfx90a;gfx94
0"
\
-D
GPU_TARGETS
=
"gfx
110
0"
\
-D
CMAKE_VERBOSE_MAKEFILE:BOOL
=
ON
\
-D
CMAKE_VERBOSE_MAKEFILE:BOOL
=
ON
\
-D
USE_BITINT_EXTENSION_INT4
=
OFF
\
-D
USE_BITINT_EXTENSION_INT4
=
OFF
\
${
MY_PROJECT_SOURCE
}
${
MY_PROJECT_SOURCE
}
...
...
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