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
253f942b
Commit
253f942b
authored
Sep 22, 2023
by
Umang Yadav
Browse files
changes to make it compile
parent
8f9c0243
Changes
275
Hide whitespace changes
Inline
Side-by-side
Showing
20 changed files
with
110 additions
and
7 deletions
+110
-7
include/ck/tensor_operation/gpu/device/impl/device_pool3d_fwd_ndhwc_ndhwc.hpp
...eration/gpu/device/impl/device_pool3d_fwd_ndhwc_ndhwc.hpp
+5
-0
include/ck/tensor_operation/gpu/device/impl/device_reduce_common.hpp
...tensor_operation/gpu/device/impl/device_reduce_common.hpp
+5
-0
include/ck/tensor_operation/gpu/device/impl/device_reduce_multiblock.hpp
...or_operation/gpu/device/impl/device_reduce_multiblock.hpp
+5
-0
include/ck/tensor_operation/gpu/device/impl/device_reduce_threadwise.hpp
...or_operation/gpu/device/impl/device_reduce_threadwise.hpp
+5
-0
include/ck/tensor_operation/gpu/device/impl/device_softmax_impl.hpp
.../tensor_operation/gpu/device/impl/device_softmax_impl.hpp
+8
-2
include/ck/tensor_operation/gpu/device/impl/device_sparse_embeddings_forward_layernorm.hpp
...evice/impl/device_sparse_embeddings_forward_layernorm.hpp
+5
-0
include/ck/tensor_operation/gpu/device/impl/device_splitk_contraction_multiple_d_xdl_cshuffle.hpp
...mpl/device_splitk_contraction_multiple_d_xdl_cshuffle.hpp
+10
-4
include/ck/tensor_operation/gpu/device/masking_specialization.hpp
...ck/tensor_operation/gpu/device/masking_specialization.hpp
+5
-0
include/ck/tensor_operation/gpu/device/matrix_padder.hpp
include/ck/tensor_operation/gpu/device/matrix_padder.hpp
+5
-0
include/ck/tensor_operation/gpu/device/reduction_operator_mapping.hpp
...ensor_operation/gpu/device/reduction_operator_mapping.hpp
+5
-0
include/ck/tensor_operation/gpu/device/tensor_layout.hpp
include/ck/tensor_operation/gpu/device/tensor_layout.hpp
+5
-0
include/ck/tensor_operation/gpu/device/tensor_specialization.hpp
.../ck/tensor_operation/gpu/device/tensor_specialization.hpp
+5
-0
include/ck/tensor_operation/gpu/device/welford_helper.hpp
include/ck/tensor_operation/gpu/device/welford_helper.hpp
+5
-0
include/ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp
...r_operation/gpu/element/binary_element_wise_operation.hpp
+5
-0
include/ck/tensor_operation/gpu/element/element_wise_operation.hpp
...k/tensor_operation/gpu/element/element_wise_operation.hpp
+5
-0
include/ck/tensor_operation/gpu/element/quantization_operation.hpp
...k/tensor_operation/gpu/element/quantization_operation.hpp
+5
-0
include/ck/tensor_operation/gpu/element/unary_element_wise_operation.hpp
...or_operation/gpu/element/unary_element_wise_operation.hpp
+7
-1
include/ck/tensor_operation/gpu/grid/batchnorm_multiblock/gridwise_multiblock_reduce_second_half_batchnorm_backward_final.hpp
...ultiblock_reduce_second_half_batchnorm_backward_final.hpp
+5
-0
include/ck/tensor_operation/gpu/grid/batchnorm_multiblock/gridwise_multiblock_welford_first_half.hpp
...orm_multiblock/gridwise_multiblock_welford_first_half.hpp
+5
-0
include/ck/tensor_operation/gpu/grid/batchnorm_multiblock/gridwise_multiblock_welford_second_half_batchnorm_forward_final_obsolete.hpp
..._welford_second_half_batchnorm_forward_final_obsolete.hpp
+5
-0
No files found.
include/ck/tensor_operation/gpu/device/impl/device_pool3d_fwd_ndhwc_ndhwc.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -409,3 +412,5 @@ struct DevicePool3dFwd_NDHWC_NDHWC : public DevicePoolFwd<5,
}
// namespace device
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/impl/device_reduce_common.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -140,3 +143,5 @@ shuffle_tensor_dimensions(const std::array<index_t, Rank>& origLengthsStrides,
}
// namespace device
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/impl/device_reduce_multiblock.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -549,3 +552,5 @@ struct DeviceReduceMultiBlock : public DeviceReduce<InDataType,
}
// namespace device
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/impl/device_reduce_threadwise.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -391,3 +394,5 @@ struct DeviceReduceThreadWise : public DeviceReduce<InDataType,
}
// namespace device
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/impl/device_softmax_impl.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -386,8 +389,9 @@ struct DeviceSoftmaxImpl : public DeviceSoftmax<InDataType,
in_elementwise_op
,
acc_elementwise_op
);
};
static
auto
MakeInvoker
()
{
return
Invoker
{};
}
#ifndef __HIPCC_RTC__
static
auto
MakeInvoker
()
{
return
Invoker
{};
}
#endif
std
::
unique_ptr
<
BaseInvoker
>
MakeInvokerPointer
()
override
{
...
...
@@ -415,3 +419,5 @@ struct DeviceSoftmaxImpl : public DeviceSoftmax<InDataType,
}
// namespace device
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/impl/device_sparse_embeddings_forward_layernorm.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -191,3 +194,5 @@ struct DeviceSparseEmbeddingsForwardLayernorm : public BaseOperator
}
// namespace device
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/impl/device_splitk_contraction_multiple_d_xdl_cshuffle.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -939,7 +942,7 @@ struct DeviceSplitKContractionMultipleD_Xdl_CShuffle
return
Run
(
*
dynamic_cast
<
const
Argument
*>
(
p_arg
),
stream_config
);
}
};
#ifndef __HIPCC_RTC__
static
bool
IsSupportedArgument
(
const
Argument
&
arg
)
{
if
(
!
ck
::
is_xdl_supported
())
...
...
@@ -1037,7 +1040,7 @@ struct DeviceSplitKContractionMultipleD_Xdl_CShuffle
{
return
IsSupportedArgument
(
*
dynamic_cast
<
const
Argument
*>
(
p_arg
));
}
#endif
static
auto
MakeArgument
(
const
void
*
p_a
,
const
void
*
p_b
,
...
...
@@ -1074,8 +1077,9 @@ struct DeviceSplitKContractionMultipleD_Xdl_CShuffle
split_k
};
}
static
auto
MakeInvoker
()
{
return
Invoker
{};
}
#ifndef __HIPCC_RTC__
static
auto
MakeInvoker
()
{
return
Invoker
{};
}
#endif
// polymorphic
std
::
unique_ptr
<
BaseArgument
>
MakeArgumentPointer
(
const
void
*
p_a
,
...
...
@@ -1149,3 +1153,5 @@ struct DeviceSplitKContractionMultipleD_Xdl_CShuffle
}
// namespace device
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/masking_specialization.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -80,3 +83,5 @@ struct C0MatrixMask_impl
}
// namespace device
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/matrix_padder.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -380,3 +383,5 @@ struct MatrixPadder_v2
}
// namespace device
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/reduction_operator_mapping.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -184,3 +187,5 @@ struct reduce_unary_operator<ReduceTensorOp::NORM2, false, true>
};
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/tensor_layout.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -415,3 +418,5 @@ std::ostream& operator<<(std::ostream& os, const Layout&)
}
// namespace tensor_layout
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/tensor_specialization.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -26,3 +29,5 @@ inline std::string getTensorSpecializationString(const TensorSpecialization& s)
}
// namespace device
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/device/welford_helper.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -87,3 +90,5 @@ struct GetReduceCountPerThreadForMultiblockWelford
}
// namespace device
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -323,3 +326,5 @@ struct AddFastGelu
}
// namespace element_wise
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/element/element_wise_operation.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -392,3 +395,5 @@ struct UnaryTypeConvert<ck::bhalf_t, float>
}
// namespace element_wise
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/element/quantization_operation.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
#pragma once
#include "ck/utility/data_type.hpp"
...
...
@@ -284,3 +287,5 @@ struct Add_Mul2_Activation_Mul_Clamp
}
// namespace element_wise
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/element/unary_element_wise_operation.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -288,6 +291,7 @@ struct FastGelu
template
<
typename
Y
,
typename
X
>
__device__
void
operator
()(
Y
&
y
,
const
X
&
x
)
const
;
#ifndef __HIPCC_RTC__
template
<
>
__host__
void
operator
()
<
float
,
float
>
(
float
&
y
,
const
float
&
x
)
const
{
...
...
@@ -297,7 +301,7 @@ struct FastGelu
y
=
x
*
cdf
;
}
#endif
// device code, use lower precision "__expf" and "rcp"
template
<
>
__device__
void
operator
()
<
float
,
float
>
(
float
&
y
,
const
float
&
x
)
const
...
...
@@ -422,3 +426,5 @@ struct Swish
}
// namespace element_wise
}
// namespace tensor_operation
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/grid/batchnorm_multiblock/gridwise_multiblock_reduce_second_half_batchnorm_backward_final.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -496,3 +499,5 @@ struct GridwiseReduceSecondHalfBatchNormBackwardFinal
};
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/grid/batchnorm_multiblock/gridwise_multiblock_welford_first_half.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -259,3 +262,5 @@ struct GridwiseMultiblockWelfordFirstHalf
};
}
// namespace ck
#pragma clang diagnostic pop
include/ck/tensor_operation/gpu/grid/batchnorm_multiblock/gridwise_multiblock_welford_second_half_batchnorm_forward_final_obsolete.hpp
View file @
253f942b
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Weverything"
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
...
...
@@ -568,3 +571,5 @@ struct GridwiseWelfordSecondHalfBatchNormForwardFinal
};
}
// namespace ck
#pragma clang diagnostic pop
Prev
1
…
4
5
6
7
8
9
10
11
12
…
14
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