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_ROCM
Commits
a634647d
Commit
a634647d
authored
Nov 14, 2024
by
Andriy Roshchenko
Browse files
Cleanup
parent
ab9f0eda
Changes
5
Hide whitespace changes
Inline
Side-by-side
Showing
5 changed files
with
0 additions
and
275 deletions
+0
-275
CMakePresets.json
CMakePresets.json
+0
-189
include/ck/utility/data_type.hpp
include/ck/utility/data_type.hpp
+0
-5
script/fp8_tests.list
script/fp8_tests.list
+0
-46
script/make_fp8_tests.sh
script/make_fp8_tests.sh
+0
-32
script/run_fp8_gemm_universal_tests.sh
script/run_fp8_gemm_universal_tests.sh
+0
-3
No files found.
CMakePresets.json
deleted
100644 → 0
View file @
ab9f0eda
{
"version"
:
3
,
"configurePresets"
:
[
{
"name"
:
"linux-debug"
,
"displayName"
:
"Linux Debug"
,
"hidden"
:
true
,
"generator"
:
"Unix Makefiles"
,
"binaryDir"
:
"${sourceDir}/build/${presetName}"
,
"installDir"
:
"${sourceDir}/build/install/${presetName}"
,
"environment"
:
{
"MY_ENVIRONMENT_VARIABLE"
:
"NONE"
,
"PATH"
:
"/usr/local/.cargo/bin:$penv{PATH}"
,
"SCCACHE_IDLE_TIMEOUT"
:
"11000"
},
"cacheVariables"
:
{
"CMAKE_BUILD_TYPE"
:
"Debug"
,
"CMAKE_EXPORT_COMPILE_COMMANDS"
:
"ON"
,
"BUILD_DEV"
:
"ON"
,
"CMAKE_CXX_COMPILER"
:
"/opt/rocm-6.3.0/llvm/bin/clang++"
,
"CMAKE_PREFIX_PATH"
:
"/opt/rocm-6.3.0"
,
"CMAKE_CXX_COMPILER_LAUNCHER"
:
"sccache"
,
"CMAKE_C_COMPILER_LAUNCHER"
:
"sccache"
},
"condition"
:
{
"type"
:
"equals"
,
"lhs"
:
"${hostSystemName}"
,
"rhs"
:
"Linux"
}
},
{
"name"
:
"MI355-debug"
,
"displayName"
:
"MI355 Debug"
,
"inherits"
:
"linux-debug"
,
"description"
:
"Development Environment for MI355."
,
"cacheVariables"
:
{
"GPU_TARGETS"
:
"gfx950"
,
"CMAKE_BUILD_TYPE"
:
"Debug"
,
"CMAKE_CXX_FLAGS"
:
"-O0 -ggdb"
}
},
{
"name"
:
"MI355-release"
,
"displayName"
:
"MI355 Release"
,
"inherits"
:
"linux-debug"
,
"cacheVariables"
:
{
"GPU_TARGETS"
:
"gfx950"
,
"CMAKE_BUILD_TYPE"
:
"Release"
,
"CMAKE_CXX_FLAGS"
:
"-O3"
}
},
{
"name"
:
"MI300X-release"
,
"displayName"
:
"MI300X Release"
,
"inherits"
:
"linux-debug"
,
"cacheVariables"
:
{
"GPU_TARGETS"
:
"gfx942"
,
"CMAKE_BUILD_TYPE"
:
"Release"
,
"CMAKE_CXX_FLAGS"
:
"-O3"
}
},
{
"name"
:
"MI250-release"
,
"displayName"
:
"MI250 Release"
,
"inherits"
:
"linux-debug"
,
"cacheVariables"
:
{
"GPU_TARGETS"
:
"gfx90a"
,
"CMAKE_BUILD_TYPE"
:
"Release"
,
"CMAKE_CXX_FLAGS"
:
"-O3"
,
"CK_USE_FP8_ON_UNSUPPORTED_ARCH"
:
"ON"
}
},
{
"name"
:
"MI250-debug"
,
"displayName"
:
"MI250 Debug"
,
"inherits"
:
"linux-debug"
,
"cacheVariables"
:
{
"GPU_TARGETS"
:
"gfx90a"
,
"CMAKE_BUILD_TYPE"
:
"Debug"
,
"CMAKE_CXX_FLAGS"
:
"-O0 -ggdb"
,
"CK_USE_FP8_ON_UNSUPPORTED_ARCH"
:
"ON"
}
},
{
"name"
:
"RX7800-release"
,
"displayName"
:
"RX7800 Release"
,
"inherits"
:
"linux-debug"
,
"cacheVariables"
:
{
"GPU_TARGETS"
:
"gfx1101"
,
"DL_KERNELS"
:
"ON"
,
"CMAKE_BUILD_TYPE"
:
"Release"
,
"CMAKE_CXX_FLAGS"
:
"-O3"
}
},
{
"name"
:
"RX7800-debug"
,
"displayName"
:
"RX7800 Debug"
,
"inherits"
:
"linux-debug"
,
"cacheVariables"
:
{
"GPU_TARGETS"
:
"gfx1101"
,
"DL_KERNELS"
:
"ON"
,
"CMAKE_BUILD_TYPE"
:
"Debug"
,
"CMAKE_CXX_FLAGS"
:
"-O0 -ggdb"
}
}
],
"buildPresets"
:
[
{
"name"
:
"Debug"
,
"hidden"
:
true
,
"configuration"
:
"Debug"
},
{
"name"
:
"Release"
,
"hidden"
:
true
,
"configuration"
:
"Release"
},
{
"name"
:
"MI355-debug"
,
"displayName"
:
"MI355"
,
"configurePreset"
:
"MI355-debug"
,
"description"
:
"Build Environment for MI355 Debug."
,
"inherits"
:
[
"Debug"
],
"jobs"
:
128
},
{
"name"
:
"MI355-release"
,
"displayName"
:
"MI355"
,
"configurePreset"
:
"MI355-release"
,
"description"
:
"Build Environment for MI355 Release."
,
"inherits"
:
[
"Release"
],
"jobs"
:
128
},
{
"name"
:
"MI300X-release"
,
"displayName"
:
"MI300X"
,
"configurePreset"
:
"MI300X-release"
,
"description"
:
"Build Environment for MI300X Release."
,
"inherits"
:
[
"Release"
],
"jobs"
:
128
},
{
"name"
:
"MI250-release"
,
"displayName"
:
"MI250"
,
"configurePreset"
:
"MI250-release"
,
"description"
:
"Build Environment for MI250 Release."
,
"inherits"
:
[
"Release"
],
"jobs"
:
128
},
{
"name"
:
"MI250-debug"
,
"displayName"
:
"MI250"
,
"configurePreset"
:
"MI250-debug"
,
"description"
:
"Build Environment for MI250 Debug."
,
"inherits"
:
[
"Debug"
],
"jobs"
:
128
},
{
"name"
:
"RX7800-release"
,
"displayName"
:
"RX7800"
,
"configurePreset"
:
"RX7800-release"
,
"description"
:
"Build Environment for RX7800 Release."
,
"inherits"
:
[
"Release"
],
"jobs"
:
128
},
{
"name"
:
"RX7800-debug"
,
"displayName"
:
"RX7800"
,
"configurePreset"
:
"RX7800-debug"
,
"description"
:
"Build Environment for RX7800 Debug."
,
"inherits"
:
[
"Debug"
],
"jobs"
:
128
}
]
}
include/ck/utility/data_type.hpp
View file @
a634647d
...
@@ -1069,7 +1069,6 @@ struct vector_type<T, 1, typename std::enable_if_t<!is_native_type<T>()>>
...
@@ -1069,7 +1069,6 @@ struct vector_type<T, 1, typename std::enable_if_t<!is_native_type<T>()>>
d1_t
d1_
;
d1_t
d1_
;
StaticallyIndexedArray
<
d1_t
,
1
>
d1x1_
;
StaticallyIndexedArray
<
d1_t
,
1
>
d1x1_
;
d1_nnv_t
d1_nnv_
;
d1_nnv_t
d1_nnv_
;
// StaticallyIndexedArray<d1_nnv_t, 1> d1nnvx1_;
}
data_
;
}
data_
;
__host__
__device__
constexpr
vector_type
()
:
data_
{
d1_t
{}}
{}
__host__
__device__
constexpr
vector_type
()
:
data_
{
d1_t
{}}
{}
...
@@ -1122,7 +1121,6 @@ struct vector_type<T, 2, typename std::enable_if_t<!is_native_type<T>()>>
...
@@ -1122,7 +1121,6 @@ struct vector_type<T, 2, typename std::enable_if_t<!is_native_type<T>()>>
{
{
d2_t
d2_
;
d2_t
d2_
;
StaticallyIndexedArray
<
d1_t
,
2
>
d1x2_
;
StaticallyIndexedArray
<
d1_t
,
2
>
d1x2_
;
// StaticallyIndexedArray<d1_nnv_t, 2> d1nnvx2_;
StaticallyIndexedArray
<
d2_t
,
1
>
d2x1_
;
StaticallyIndexedArray
<
d2_t
,
1
>
d2x1_
;
}
data_
;
}
data_
;
...
@@ -1187,7 +1185,6 @@ struct vector_type<T, 4, typename std::enable_if_t<!is_native_type<T>()>>
...
@@ -1187,7 +1185,6 @@ struct vector_type<T, 4, typename std::enable_if_t<!is_native_type<T>()>>
{
{
d4_t
d4_
;
d4_t
d4_
;
StaticallyIndexedArray
<
d1_t
,
4
>
d1x4_
;
StaticallyIndexedArray
<
d1_t
,
4
>
d1x4_
;
// StaticallyIndexedArray<d1_nnv_t, 4> d1nnvx4_;
StaticallyIndexedArray
<
d2_t
,
2
>
d2x2_
;
StaticallyIndexedArray
<
d2_t
,
2
>
d2x2_
;
StaticallyIndexedArray
<
d4_t
,
1
>
d4x1_
;
StaticallyIndexedArray
<
d4_t
,
1
>
d4x1_
;
}
data_
;
}
data_
;
...
@@ -1262,7 +1259,6 @@ struct vector_type<T, 8, typename std::enable_if_t<!is_native_type<T>()>>
...
@@ -1262,7 +1259,6 @@ struct vector_type<T, 8, typename std::enable_if_t<!is_native_type<T>()>>
{
{
d8_t
d8_
;
d8_t
d8_
;
StaticallyIndexedArray
<
d1_t
,
8
>
d1x8_
;
StaticallyIndexedArray
<
d1_t
,
8
>
d1x8_
;
// StaticallyIndexedArray<d1_nnv_t, 8> d1nnvx8_;
StaticallyIndexedArray
<
d2_t
,
4
>
d2x4_
;
StaticallyIndexedArray
<
d2_t
,
4
>
d2x4_
;
StaticallyIndexedArray
<
d4_t
,
2
>
d4x2_
;
StaticallyIndexedArray
<
d4_t
,
2
>
d4x2_
;
StaticallyIndexedArray
<
d8_t
,
1
>
d8x1_
;
StaticallyIndexedArray
<
d8_t
,
1
>
d8x1_
;
...
@@ -1349,7 +1345,6 @@ struct vector_type<T, 16, typename std::enable_if_t<!is_native_type<T>()>>
...
@@ -1349,7 +1345,6 @@ struct vector_type<T, 16, typename std::enable_if_t<!is_native_type<T>()>>
{
{
d16_t
d16_
;
d16_t
d16_
;
StaticallyIndexedArray
<
d1_t
,
16
>
d1x16_
;
StaticallyIndexedArray
<
d1_t
,
16
>
d1x16_
;
// StaticallyIndexedArray<d1_nnv_t, 16> d1nnvx16_;
StaticallyIndexedArray
<
d2_t
,
8
>
d2x8_
;
StaticallyIndexedArray
<
d2_t
,
8
>
d2x8_
;
StaticallyIndexedArray
<
d4_t
,
4
>
d4x4_
;
StaticallyIndexedArray
<
d4_t
,
4
>
d4x4_
;
StaticallyIndexedArray
<
d8_t
,
2
>
d8x2_
;
StaticallyIndexedArray
<
d8_t
,
2
>
d8x2_
;
...
...
script/fp8_tests.list
deleted
100644 → 0
View file @
ab9f0eda
## Captured by CTest for MI300X `ctest -N -R "fp8|bf8"`
example_gemm_xdl_fp8_v3
example_gemm_xdl_fp16_fp8_v3
example_gemm_xdl_fp8
example_gemm_xdl_fp8_bf8
example_gemm_xdl_fp16_fp8
example_convnd_fwd_xdl_fp8
example_convnd_fwd_xdl_bf8
example_convnd_fwd_xdl_fp16_comp_fp8
example_convnd_fwd_xdl_fp8_bf8
example_convnd_fwd_xdl_bf8_fp8
example_grouped_gemm_xdl_fixed_nk_fp16_fp8
example_grouped_conv_bwd_weight_xdl_fp16_comp_bf8_fp8
example_splitK_gemm_xdl_fp16_fp8
example_grouped_conv_bwd_data_xdl_fp16_comp_bf8_fp8
elementwise_scale_permute_amax_2D_fp16_fp8
example_convnd_fwd_xdl_convinvscale_fp8
example_convnd_fwd_xdl_convscale_fp8
example_convnd_fwd_xdl_convscale_bf8
example_convnd_fwd_xdl_convscale_fp8_bf8
example_convnd_fwd_xdl_convscale_bf8_fp8
example_convnd_fwd_xdl_convscale_relu_fp8
example_convnd_fwd_xdl_convscale_add_fp8
example_convnd_fwd_xdl_convscale_relu_amax_fp8
example_convnd_fwd_xdl_convscale_amax_fp8
example_gemm_multiply_multiply_xdl_fp8
example_gemm_multiply_multiply_xdl_fp8_ab_scale
test_fp8_fnuz
test_bf8_fnuz
test_fp8_ocp
test_bf8_ocp
## Found by searching for `f8` in the `example` folder
example_gemm_add_add_xdl_fp16
## Found by searching for `f8` in the `test` folder
test_custom_type
### Takes too much time to run on the emulator
#### test_gemm_universal
### test_pool
test_avg_pool3d_fwd
test_max_pool3d_fwd
test_avg_pool2d_bwd
test_max_pool2d_bwd
test_avg_pool2d_fwd
test_max_pool2d_fwd
## Total Tests: 39 (1 - disabled {test_gemm_universal}, 2 - not executed {fnuz})
script/make_fp8_tests.sh
deleted
100755 → 0
View file @
ab9f0eda
make
-j
15
\
example_gemm_xdl_fp8_v3
\
example_gemm_xdl_fp16_fp8_v3
\
example_gemm_xdl_fp8
\
example_gemm_xdl_fp8_bf8
\
example_gemm_xdl_fp16_fp8
\
example_convnd_fwd_xdl_fp8
\
example_convnd_fwd_xdl_bf8
\
example_convnd_fwd_xdl_fp16_comp_fp8
\
example_convnd_fwd_xdl_fp8_bf8
\
example_convnd_fwd_xdl_bf8_fp8
\
example_grouped_gemm_xdl_fixed_nk_fp16_fp8
\
example_grouped_conv_bwd_weight_xdl_fp16_comp_bf8_fp8
\
example_splitK_gemm_xdl_fp16_fp8
\
example_grouped_conv_bwd_data_xdl_fp16_comp_bf8_fp8
\
elementwise_scale_permute_amax_2D_fp16_fp8
\
example_convnd_fwd_xdl_convinvscale_fp8
\
example_convnd_fwd_xdl_convscale_fp8
\
example_convnd_fwd_xdl_convscale_bf8
\
example_convnd_fwd_xdl_convscale_fp8_bf8
\
example_convnd_fwd_xdl_convscale_bf8_fp8
\
example_convnd_fwd_xdl_convscale_relu_fp8
\
example_convnd_fwd_xdl_convscale_add_fp8
\
example_convnd_fwd_xdl_convscale_relu_amax_fp8
\
example_convnd_fwd_xdl_convscale_amax_fp8
\
example_gemm_multiply_multiply_xdl_fp8
\
example_gemm_multiply_multiply_xdl_fp8_ab_scale
\
test_fp8
\
example_gemm_add_add_xdl_fp16
\
test_custom_type
\
test_gemm_universal
\
test_pool
\ No newline at end of file
script/run_fp8_gemm_universal_tests.sh
deleted
100755 → 0
View file @
ab9f0eda
GEMM_EXE
=
./bin/test_gemm_universal
$GEMM_EXE
--gtest_filter
=
TestGemmUniversal_MK_KN/1.SmallM:TestGemmUniversal_MK_KN/1.MidLargeM:TestGemmUniversal_MK_KN/1.PaddK:TestGemmUniversal_MK_KN/1.Regular:TestGemmUniversal_MK_KN/2.SmallM:TestGemmUniversal_MK_KN/2.MidLargeM:TestGemmUniversal_MK_KN/2.PaddK:TestGemmUniversal_MK_KN/2.Regular:TestGemmUniversal_MK_KN/3.SmallM:TestGemmUniversal_MK_KN/3.MidLargeM:TestGemmUniversal_MK_KN/3.PaddK:TestGemmUniversal_MK_KN/3.Regular:TestGemmUniversal_MK_NK/1.SmallM:TestGemmUniversal_MK_NK/1.MidLargeM:TestGemmUniversal_MK_NK/1.PaddK:TestGemmUniversal_MK_NK/1.Regular:TestGemmUniversal_MK_NK/2.SmallM:TestGemmUniversal_MK_NK/2.MidLargeM:TestGemmUniversal_MK_NK/2.PaddK:TestGemmUniversal_MK_NK/2.Regular:TestGemmUniversal_MK_NK/3.SmallM:TestGemmUniversal_MK_NK/3.MidLargeM:TestGemmUniversal_MK_NK/3.PaddK:TestGemmUniversal_MK_NK/3.Regular
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