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
cba8f7f2
Commit
cba8f7f2
authored
Jun 26, 2022
by
Anthony Chang
Browse files
Merge remote-tracking branch 'upstream/develop' into gemm-layernorm-4
parents
cc50b687
b653c5eb
Changes
583
Hide whitespace changes
Inline
Side-by-side
Showing
20 changed files
with
92 additions
and
42 deletions
+92
-42
include/ck/problem_transform/transform_backward_weight_convolution_into_gemm_v4r4r4_atomic_nhwc_kyxc_nhwk.hpp
...ht_convolution_into_gemm_v4r4r4_atomic_nhwc_kyxc_nhwk.hpp
+3
-0
include/ck/problem_transform/transform_backward_weight_convolution_into_gemm_v4r4r4_nhwc_kyxc_nhwk.hpp
...rd_weight_convolution_into_gemm_v4r4r4_nhwc_kyxc_nhwk.hpp
+3
-0
include/ck/problem_transform/transform_backward_weight_convolution_into_gemm_v4r4r5_nhwc_kyxc_nhwk.hpp
...rd_weight_convolution_into_gemm_v4r4r5_nhwc_kyxc_nhwk.hpp
+3
-0
include/ck/problem_transform/transform_forward_convolution3d_into_gemm_v4r4r4_ndhwc_kzyxc_ndhwk.hpp
...ward_convolution3d_into_gemm_v4r4r4_ndhwc_kzyxc_ndhwk.hpp
+3
-0
include/ck/problem_transform/transform_forward_convolution_into_gemm_v4r4_nchw_kcyx_nkhw.hpp
...orm_forward_convolution_into_gemm_v4r4_nchw_kcyx_nkhw.hpp
+3
-0
include/ck/problem_transform/transform_forward_convolution_into_gemm_v4r4_nhwc_kyxc_nhwk.hpp
...orm_forward_convolution_into_gemm_v4r4_nhwc_kyxc_nhwk.hpp
+3
-0
include/ck/problem_transform/transform_forward_convolution_into_gemm_v4r4r2_nchw_kcyx_nkhw.hpp
...m_forward_convolution_into_gemm_v4r4r2_nchw_kcyx_nkhw.hpp
+3
-0
include/ck/problem_transform/transform_forward_convolution_into_gemm_v4r4r2_nhwc_kyxc_nhwk.hpp
...m_forward_convolution_into_gemm_v4r4r2_nhwc_kyxc_nhwk.hpp
+3
-0
include/ck/problem_transform/transform_forward_convolution_into_gemm_v4r4r4_nhwc_kyxc_nhwk.hpp
...m_forward_convolution_into_gemm_v4r4r4_nhwc_kyxc_nhwk.hpp
+3
-0
include/ck/problem_transform/transform_forward_convolution_into_gemm_v6r1_nchw_kcyx_nkhw.hpp
...orm_forward_convolution_into_gemm_v6r1_nchw_kcyx_nkhw.hpp
+3
-0
include/ck/stream_config.hpp
include/ck/stream_config.hpp
+3
-0
include/ck/tensor/static_tensor.hpp
include/ck/tensor/static_tensor.hpp
+3
-0
include/ck/tensor_description/cluster_descriptor.hpp
include/ck/tensor_description/cluster_descriptor.hpp
+6
-5
include/ck/tensor_description/multi_index_transform.hpp
include/ck/tensor_description/multi_index_transform.hpp
+6
-5
include/ck/tensor_description/multi_index_transform_helper.hpp
...de/ck/tensor_description/multi_index_transform_helper.hpp
+6
-5
include/ck/tensor_description/tensor_adaptor.hpp
include/ck/tensor_description/tensor_adaptor.hpp
+7
-6
include/ck/tensor_description/tensor_descriptor.hpp
include/ck/tensor_description/tensor_descriptor.hpp
+6
-5
include/ck/tensor_description/tensor_descriptor_helper.hpp
include/ck/tensor_description/tensor_descriptor_helper.hpp
+7
-3
include/ck/tensor_description/tensor_space_filling_curve.hpp
include/ck/tensor_description/tensor_space_filling_curve.hpp
+10
-9
include/ck/tensor_operation/gpu/block/blockwise_gemm_dl_v2r3.hpp
.../ck/tensor_operation/gpu/block/blockwise_gemm_dl_v2r3.hpp
+8
-4
No files found.
include/ck/problem_transform/transform_backward_weight_convolution_into_gemm_v4r4r4_atomic_nhwc_kyxc_nhwk.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R4_ATOMIC_NHWC_KYXC_NHWK_HPP
#ifndef CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R4_ATOMIC_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R4_ATOMIC_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R4_ATOMIC_NHWC_KYXC_NHWK_HPP
...
...
include/ck/problem_transform/transform_backward_weight_convolution_into_gemm_v4r4r4_nhwc_kyxc_nhwk.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
#ifndef CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
...
...
include/ck/problem_transform/transform_backward_weight_convolution_into_gemm_v4r4r5_nhwc_kyxc_nhwk.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R5_NHWC_KYXC_NHWK_HPP
#ifndef CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R5_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R5_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_BACKWARD_WEIGHT_CONVOLUTION_INTO_GEMM_V4R4R5_NHWC_KYXC_NHWK_HPP
...
...
include/ck/problem_transform/transform_forward_convolution3d_into_gemm_v4r4r4_ndhwc_kzyxc_ndhwk.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION3D_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION3D_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION3D_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION3D_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
...
...
include/ck/problem_transform/transform_forward_convolution_into_gemm_v4r4_nchw_kcyx_nkhw.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4_NCHW_KCYX_NKHW_HPP
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4_NCHW_KCYX_NKHW_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4_NCHW_KCYX_NKHW_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4_NCHW_KCYX_NKHW_HPP
...
...
include/ck/problem_transform/transform_forward_convolution_into_gemm_v4r4_nhwc_kyxc_nhwk.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4_NHWC_KYXC_NHWK_HPP
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4_NHWC_KYXC_NHWK_HPP
...
...
include/ck/problem_transform/transform_forward_convolution_into_gemm_v4r4r2_nchw_kcyx_nkhw.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R2_NCHW_KCYX_NKHW_HPP
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R2_NCHW_KCYX_NKHW_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R2_NCHW_KCYX_NKHW_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R2_NCHW_KCYX_NKHW_HPP
...
...
include/ck/problem_transform/transform_forward_convolution_into_gemm_v4r4r2_nhwc_kyxc_nhwk.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R2_NHWC_KYXC_NHWK_HPP
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R2_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R2_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R2_NHWC_KYXC_NHWK_HPP
...
...
include/ck/problem_transform/transform_forward_convolution_into_gemm_v4r4r4_nhwc_kyxc_nhwk.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_GEMM_V4R4R4_NHWC_KYXC_NHWK_HPP
...
...
include/ck/problem_transform/transform_forward_convolution_into_gemm_v6r1_nchw_kcyx_nkhw.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_CONTRACTION_V6R1_NCHW_KCYX_NKHW_HPP
#ifndef CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_CONTRACTION_V6R1_NCHW_KCYX_NKHW_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_CONTRACTION_V6R1_NCHW_KCYX_NKHW_HPP
#define CK_TRANSFORM_FORWARD_CONVOLUTION_INTO_CONTRACTION_V6R1_NCHW_KCYX_NKHW_HPP
...
...
include/ck/stream_config.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#pragma once
#include <hip/hip_runtime.h>
#include <hip/hip_runtime.h>
...
...
include/ck/tensor/static_tensor.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_STATIC_TENSOR_HPP
#ifndef CK_STATIC_TENSOR_HPP
#define CK_STATIC_TENSOR_HPP
#define CK_STATIC_TENSOR_HPP
...
...
include/ck/tensor_description/cluster_descriptor.hpp
View file @
cba8f7f2
#ifndef CK_CLUSTER_DESCRIPTOR_HPP
// SPDX-License-Identifier: MIT
#define CK_CLUSTER_DESCRIPTOR_HPP
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#include "common_header.hpp"
#pragma once
#include "tensor_adaptor.hpp"
#include "ck/utility/common_header.hpp"
#include "ck/tensor_description/tensor_adaptor.hpp"
namespace
ck
{
namespace
ck
{
...
@@ -30,4 +32,3 @@ __host__ __device__ constexpr auto make_cluster_descriptor(
...
@@ -30,4 +32,3 @@ __host__ __device__ constexpr auto make_cluster_descriptor(
}
}
}
// namespace ck
}
// namespace ck
#endif
include/ck/tensor_description/multi_index_transform.hpp
View file @
cba8f7f2
#ifndef CK_MULTI_INDEX_TRANSFORM_HPP
// SPDX-License-Identifier: MIT
#define CK_MULTI_INDEX_TRANSFORM_HPP
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#include "common_header.hpp"
#pragma once
#include "multi_index.hpp"
#include "ck/utility/common_header.hpp"
#include "ck/utility/multi_index.hpp"
namespace
ck
{
namespace
ck
{
...
@@ -1950,4 +1952,3 @@ struct Modulo
...
@@ -1950,4 +1952,3 @@ struct Modulo
}
}
};
};
}
// namespace ck
}
// namespace ck
#endif
include/ck/tensor_description/multi_index_transform_helper.hpp
View file @
cba8f7f2
#ifndef CK_MULTI_INDEX_TRANSFORM_HELPER_HPP
// SPDX-License-Identifier: MIT
#define CK_MULTI_INDEX_TRANSFORM_HELPER_HPP
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#include "common_header.hpp"
#pragma once
#include "multi_index_transform.hpp"
#include "ck/utility/common_header.hpp"
#include "ck/tensor_description/multi_index_transform.hpp"
namespace
ck
{
namespace
ck
{
...
@@ -126,4 +128,3 @@ __host__ __device__ constexpr auto make_modulo_transform(const Modulus& modulus,
...
@@ -126,4 +128,3 @@ __host__ __device__ constexpr auto make_modulo_transform(const Modulus& modulus,
return
Modulo
<
Modulus
,
UpLength
>
{
modulus
,
up_length
};
return
Modulo
<
Modulus
,
UpLength
>
{
modulus
,
up_length
};
}
}
}
// namespace ck
}
// namespace ck
#endif
include/ck/tensor_description/tensor_adaptor.hpp
View file @
cba8f7f2
#ifndef CK_TENSOR_ADAPTOR_HPP
// SPDX-License-Identifier: MIT
#define CK_TENSOR_ADAPTOR_HPP
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#include "common_header.hpp"
#pragma once
#include "tensor_descriptor.hpp"
#include "tensor_descriptor_helper.hpp"
#include "ck/utility/common_header.hpp"
#include "ck/tensor_description/tensor_descriptor.hpp"
#include "ck/tensor_description/tensor_descriptor_helper.hpp"
namespace
ck
{
namespace
ck
{
...
@@ -478,4 +480,3 @@ __host__ __device__ constexpr auto chain_tensor_adaptors(const X& x, const Xs&..
...
@@ -478,4 +480,3 @@ __host__ __device__ constexpr auto chain_tensor_adaptors(const X& x, const Xs&..
}
}
}
// namespace ck
}
// namespace ck
#endif
include/ck/tensor_description/tensor_descriptor.hpp
View file @
cba8f7f2
#ifndef CK_TENSOR_DESCRIPTOR_HPP
// SPDX-License-Identifier: MIT
#define CK_TENSOR_DESCRIPTOR_HPP
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#include "common_header.hpp"
#pragma once
#include "multi_index_transform.hpp"
#include "ck/utility/common_header.hpp"
#include "ck/tensor_description/multi_index_transform.hpp"
namespace
ck
{
namespace
ck
{
...
@@ -604,4 +606,3 @@ using TensorCoordinateStep_t = decltype(make_tensor_coordinate_step(
...
@@ -604,4 +606,3 @@ using TensorCoordinateStep_t = decltype(make_tensor_coordinate_step(
TensorDesc
{},
MultiIndex
<
remove_cvref_t
<
TensorDesc
>::
GetNumOfDimension
()
>
{}));
TensorDesc
{},
MultiIndex
<
remove_cvref_t
<
TensorDesc
>::
GetNumOfDimension
()
>
{}));
}
// namespace ck
}
// namespace ck
#endif
include/ck/tensor_description/tensor_descriptor_helper.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#pragma once
#include "common_header.hpp"
#include "tensor_descriptor.hpp"
#include "ck/utility/common_header.hpp"
#include "multi_index_transform_helper.hpp"
#include "ck/tensor_description/tensor_descriptor.hpp"
#include "ck/tensor_description/multi_index_transform_helper.hpp"
namespace
ck
{
namespace
ck
{
...
...
include/ck/
utility
/tensor_space_filling_curve.hpp
→
include/ck/
tensor_description
/tensor_space_filling_curve.hpp
View file @
cba8f7f2
#ifndef TENSOR_SPACE_FILLING_CURVE_HPP
// SPDX-License-Identifier: MIT
#define TENSOR_SPACE_FILLING_CURVE_HPP
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#include "math.hpp"
#pragma once
#include "sequence.hpp"
#include "sequence_helper.hpp"
#include "ck/utility/math.hpp"
#include "tensor_adaptor.hpp"
#include "ck/utility/sequence.hpp"
#include "statically_indexed_array_multi_index.hpp"
#include "ck/utility/sequence_helper.hpp"
#include "tuple_helper.hpp"
#include "ck/utility/statically_indexed_array_multi_index.hpp"
#include "ck/utility/tuple_helper.hpp"
#include "ck/tensor_description/tensor_adaptor.hpp"
namespace
ck
{
namespace
ck
{
...
@@ -156,4 +158,3 @@ struct SpaceFillingCurve
...
@@ -156,4 +158,3 @@ struct SpaceFillingCurve
};
};
}
// namespace ck
}
// namespace ck
#endif
include/ck/tensor_operation/gpu/block/blockwise_gemm_dl_v2r3.hpp
View file @
cba8f7f2
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#pragma once
#include "common_header.hpp"
#include "tensor_adaptor.hpp"
#include "ck/utility/common_header.hpp"
#include "threadwise_tensor_slice_transfer_v4r1.hpp"
#include "ck/tensor_description/tensor_adaptor.hpp"
#include "threadwise_contraction_dl.hpp"
#include "ck/tensor_operation/gpu/thread/threadwise_tensor_slice_transfer_v4r1.hpp"
#include "ck/tensor_operation/gpu/thread/threadwise_contraction_dl.hpp"
namespace
ck
{
namespace
ck
{
...
...
Prev
1
2
3
4
5
6
7
8
…
30
Next
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