Skip to content

GitLab

Commit 76f2b6cd authored by danyao12's avatar danyao12
Browse files

merge develop to attn-train-develop-qloop

parents 9b4c780a 1ee99dca
Hide whitespace changes
Inline Side-by-side
Showing with 224 additions and 114 deletions
example/41_grouped_conv_conv_fwd/CMakeLists.txt
View file @ 76f2b6cd
add_example_executable(example_grouped_conv_conv_fwd_xdl_fp32 grouped_conv_conv_fwd_xdl_fp32.cpp) list(APPEND gpu_list1 gfx908 gfx90a gfx940 gfx941 gfx942)
add_example_executable(example_grouped_conv_conv_fwd_xdl_fp16 grouped_conv_conv_fwd_xdl_fp16.cpp) list(APPEND gpu_list2 gfx908 gfx90a)
add_example_executable(example_grouped_conv_conv_fwd_xdl_bf16 grouped_conv_conv_fwd_xdl_bf16.cpp) set(target 0)
add_example_executable(example_grouped_conv_conv_fwd_xdl_int8 grouped_conv_conv_fwd_xdl_int8.cpp) foreach(gpu IN LISTS GPU_TARGETS)
if(gpu IN_LIST gpu_list1 AND target EQUAL 0)
add_example_executable(example_grouped_conv_conv_fwd_xdl_fp32 grouped_conv_conv_fwd_xdl_fp32.cpp)
add_example_executable(example_grouped_conv_conv_fwd_xdl_fp16 grouped_conv_conv_fwd_xdl_fp16.cpp)
add_example_executable(example_grouped_conv_conv_fwd_xdl_bf16 grouped_conv_conv_fwd_xdl_bf16.cpp)
if(USE_BITINT_EXTENSION_INT4)
add_example_executable(example_grouped_conv_conv_fwd_xdl_int4 grouped_conv_conv_fwd_xdl_int4.cpp)
endif(USE_BITINT_EXTENSION_INT4)
set(target 1)
endif()
endforeach()
if(USE_BITINT_EXTENSION_INT4) if(NOT GPU_TARGETS MATCHES "gfx94" AND NOT GPU_TARGETS MATCHES "gfx1")
add_example_executable(example_grouped_conv_conv_fwd_xdl_int4 grouped_conv_conv_fwd_xdl_int4.cpp) add_example_executable(example_grouped_conv_conv_fwd_xdl_int8 grouped_conv_conv_fwd_xdl_int8.cpp)
endif(USE_BITINT_EXTENSION_INT4) endif()
example/41_grouped_conv_conv_fwd/grouped_conv_conv_fwd_xdl_bf16.cpp
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT // SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved. // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#include <cstdlib> #include <cstdlib>
#include <iostream> #include <iostream>
......
example/41_grouped_conv_conv_fwd/grouped_conv_conv_fwd_xdl_fp16.cpp
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT // SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved. // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#include <cstdlib> #include <cstdlib>
#include <iostream> #include <iostream>
......
example/41_grouped_conv_conv_fwd/grouped_conv_conv_fwd_xdl_fp32.cpp
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT // SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved. // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#include <cstdlib> #include <cstdlib>
#include <iostream> #include <iostream>
......
example/41_grouped_conv_conv_fwd/grouped_conv_conv_fwd_xdl_int4.cpp
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT // SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved. // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#ifndef CK_EXPERIMENTAL_BIT_INT_EXTENSION_INT4 #ifndef CK_EXPERIMENTAL_BIT_INT_EXTENSION_INT4
#error Should compile this file with ck::int4_t support #error Should compile this file with ck::int4_t support
......
example/41_grouped_conv_conv_fwd/grouped_conv_conv_fwd_xdl_int8.cpp
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT // SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved. // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#include <cstdlib> #include <cstdlib>
#include <iostream> #include <iostream>
......
example/41_grouped_conv_conv_fwd/run_grouped_conv_conv_fwd_example.inc
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT // SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved. // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#pragma once #pragma once
......
example/42_groupnorm/CMakeLists.txt
View file @ 76f2b6cd
add_example_executable(example_groupnorm_sigmoid_fp16 groupnorm_sigmoid_fp16.cpp) add_example_executable(example_groupnorm_sigmoid_mul_fp16 groupnorm_sigmoid_mul_fp16.cpp)
add_example_executable(example_groupnorm_splitk_fp16 groupnorm_splitk_fp16.cpp)
add_example_executable(example_groupnorm_swish_fp16 groupnorm_swish_fp16.cpp)
example/42_groupnorm/common.hpp 0 → 100644
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#pragma once
#include <iostream>
#include <numeric>
#include <initializer_list>
#include <cstdlib>
#include <getopt.h>
#include "ck/ck.hpp"
#include "ck/utility/reduction_enums.hpp"
#include "ck/tensor_operation/gpu/device/impl/device_normalization_impl.hpp"
#include "ck/tensor_operation/gpu/device/impl/device_normalization_splitk_impl.hpp"
#include "ck/tensor_operation/gpu/device/reduction_operator_mapping.hpp"
#include "ck/library/utility/fill.hpp"
#include "ck/library/utility/check_err.hpp"
#include "ck/library/utility/device_memory.hpp"
#include "ck/library/utility/host_common_util.hpp"
#include "ck/library/utility/host_tensor.hpp"
#include "ck/library/utility/host_tensor_generator.hpp"
#include "ck/library/reference_tensor_operation/cpu/reference_groupnorm.hpp"
example/42_groupnorm/groupnorm_sigmoid_mul_fp16.cpp 0 → 100644
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#include "common.hpp"
constexpr int Rank = 5;
constexpr int NumReduceDim = 3;
using XDataType = ck::half_t;
using GammaDataType = ck::half_t;
using BetaDataType = ck::half_t;
using YDataType = ck::half_t;
using ComputeDataType = float;
struct YElementOp
{
template <typename T>
__host__ __device__ void operator()(T& y, const T& x) const
{
static_assert(ck::is_same<T, float>::value || ck::is_same<T, double>::value ||
ck::is_same<T, ck::half_t>::value,
"Data type is not supported by this operation!");
T a;
ck::tensor_operation::element_wise::Sigmoid{}(a, x);
y = x * a;
};
};
using DeviceInstance =
ck::tensor_operation::device::DeviceNormalizationImpl<XDataType,
GammaDataType,
BetaDataType,
ComputeDataType,
YDataType,
YElementOp,
Rank,
NumReduceDim,
1024, // BlockSize
1, // ClusterM
1024, // ClusterK
1, // SliceM
32, // SliceK
1, // SrcVecDim (0=M, 1=K)
2, // SrcScalarPerVector
1, // GammaVecDim (0=M, 1=K)
2, // GammaScalarPerVector
1, // BetaVecDim (0=M, 1=K)
2, // BetaScalarPerVector
2>; // OutScalarPerVector
#include "run_groupnorm_example.inc"
int main(int argc, char* argv[]) { run_groupnorm_example(argc, argv); }
example/42_groupnorm/groupnorm_splitk_fp16.cpp 0 → 100644
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#include "common.hpp"
constexpr int Rank = 5;
constexpr int NumReduceDim = 3;
using XDataType = ck::half_t;
using GammaDataType = ck::half_t;
using BetaDataType = ck::half_t;
using YDataType = ck::half_t;
using ComputeDataType = float;
using YElementOp = ck::tensor_operation::element_wise::Swish;
using DeviceInstance =
ck::tensor_operation::device::DeviceNormalizationSplitKImpl<XDataType,
GammaDataType,
BetaDataType,
ComputeDataType,
YDataType,
YElementOp,
Rank,
NumReduceDim,
256, // BlockSize
1, // ClusterM
256, // ClusterK
1, // SliceM
16, // SliceK
1, // SrcVecDim (0=M, 1=K)
2, // SrcScalarPerVector
1, // GammaVecDim (0=M, 1=K)
2, // GammaScalarPerVector
1, // BetaVecDim (0=M, 1=K)
2, // BetaScalarPerVector
2>; // OutScalarPerVector
#include "run_groupnorm_example.inc"
int main(int argc, char* argv[]) { run_groupnorm_example(argc, argv); }
example/42_groupnorm/groupnorm_swish_fp16.cpp 0 → 100644
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT
// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#include "common.hpp"
constexpr int Rank = 5;
constexpr int NumReduceDim = 3;
using XDataType = ck::half_t;
using GammaDataType = ck::half_t;
using BetaDataType = ck::half_t;
using YDataType = ck::half_t;
using ComputeDataType = float;
using YElementOp = ck::tensor_operation::element_wise::Swish;
using DeviceInstance =
ck::tensor_operation::device::DeviceNormalizationImpl<XDataType,
GammaDataType,
BetaDataType,
ComputeDataType,
YDataType,
YElementOp,
Rank,
NumReduceDim,
1024, // BlockSize
1, // ClusterM
1024, // ClusterK
1, // SliceM
32, // SliceK
1, // SrcVecDim (0=M, 1=K)
2, // SrcScalarPerVector
1, // GammaVecDim (0=M, 1=K)
2, // GammaScalarPerVector
1, // BetaVecDim (0=M, 1=K)
2, // BetaScalarPerVector
2>; // OutScalarPerVector
#include "run_groupnorm_example.inc"
int main(int argc, char* argv[]) { run_groupnorm_example(argc, argv); }
example/42_groupnorm/groupnorm_sigmoid_fp16.cpp → example/42_groupnorm/run_groupnorm_example.inc
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT // SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved. // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#include <iostream>
#include <numeric>
#include <initializer_list>
#include <cstdlib>
#include <getopt.h>
#include "ck/ck.hpp"
#include "ck/utility/reduction_enums.hpp"
#include "ck/tensor_operation/gpu/device/impl/device_normalization_impl.hpp"
#include "ck/tensor_operation/gpu/device/reduction_operator_mapping.hpp"
#include "ck/library/utility/fill.hpp"
#include "ck/library/utility/check_err.hpp"
#include "ck/library/utility/device_memory.hpp"
#include "ck/library/utility/host_common_util.hpp"
#include "ck/library/utility/host_tensor.hpp"
#include "ck/library/utility/host_tensor_generator.hpp"
#include "ck/library/reference_tensor_operation/cpu/reference_groupnorm.hpp"
constexpr int Rank = 5;
constexpr int NumReduceDim = 3;
using XDataType = ck::half_t;
using GammaDataType = ck::half_t;
using BetaDataType = ck::half_t;
using YDataType = ck::half_t;
using AccDataType = float;
struct YElementOp
{
template <typename T>
__host__ __device__ void operator()(T& y, const T& x) const
{
static_assert(ck::is_same<T, float>::value || ck::is_same<T, double>::value ||
ck::is_same<T, ck::half_t>::value,
"Data type is not supported by this operation!");
T a;
ck::tensor_operation::element_wise::Sigmoid{}(a, x); #pragma once
y = x * a; int run_groupnorm_example(int argc, char* argv[])
};
};
using DeviceInstance =
ck::tensor_operation::device::DeviceNormalizationImpl<XDataType,
GammaDataType,
BetaDataType,
AccDataType,
YDataType,
YElementOp,
Rank,
NumReduceDim,
1024, // BlockSize
1, // ClusterM
1024, // ClusterK
1, // SliceM
32, // SliceK
1, // SrcVecDim (0=M, 1=K)
2, // SrcScalarPerVector
1, // GammaVecDim (0=M, 1=K)
2, // GammaScalarPerVector
1, // BetaVecDim (0=M, 1=K)
2, // BetaScalarPerVector
2>; // OutScalarPerVector
int main(int argc, char* argv[])
{ {
ck::index_t N = 2; ck::index_t N = 32;
ck::index_t H = 32; ck::index_t H = 16;
ck::index_t W = 32; ck::index_t W = 16;
ck::index_t G = 32; ck::index_t G = 64;
ck::index_t C = 30; ck::index_t C = 128;
if(argc == 1) if(argc == 1)
{ {
...@@ -138,6 +73,10 @@ int main(int argc, char* argv[]) ...@@ -138,6 +73,10 @@ int main(int argc, char* argv[])
return 1; return 1;
}; };
size_t workspace_sz = device_instance.GetWorkSpaceSize(argument_ptr.get());
DeviceMem workspace_dev(workspace_sz);
device_instance.SetWorkSpacePointer(argument_ptr.get(), workspace_dev.GetDeviceBuffer());
auto invoker_ptr = device_instance.MakeInvokerPointer(); auto invoker_ptr = device_instance.MakeInvokerPointer();
float ave_time = invoker_ptr->Run(argument_ptr.get(), StreamConfig{nullptr, true, true}); float ave_time = invoker_ptr->Run(argument_ptr.get(), StreamConfig{nullptr, true, true});
...@@ -157,7 +96,7 @@ int main(int argc, char* argv[]) ...@@ -157,7 +96,7 @@ int main(int argc, char* argv[])
GammaDataType, GammaDataType,
BetaDataType, BetaDataType,
YDataType, YDataType,
AccDataType, ComputeDataType,
YElementOp>; YElementOp>;
ReferenceInstance ref; ReferenceInstance ref;
......
example/43_splitk_gemm_bias_e_permute/splitk_gemm_bias_e_permute_xdl_fp16.cpp
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT // SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved. // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#include <iostream> #include <iostream>
#include <numeric> #include <numeric>
...@@ -8,7 +8,7 @@ ...@@ -8,7 +8,7 @@
#include "ck/ck.hpp" #include "ck/ck.hpp"
#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp" #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
#include "ck/tensor_operation/gpu/device/device_splitk_contraction_multiple_d_xdl_cshuffle.hpp" #include "ck/tensor_operation/gpu/device/impl/device_splitk_contraction_multiple_d_xdl_cshuffle.hpp"
#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp" #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
#include "ck/library/utility/check_err.hpp" #include "ck/library/utility/check_err.hpp"
......
example/43_splitk_gemm_bias_e_permute/splitk_gemm_bias_e_permute_xdl_fp32.cpp
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT // SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved. // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#include <iostream> #include <iostream>
#include <numeric> #include <numeric>
...@@ -8,7 +8,7 @@ ...@@ -8,7 +8,7 @@
#include "ck/ck.hpp" #include "ck/ck.hpp"
#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp" #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
#include "ck/tensor_operation/gpu/device/device_splitk_contraction_multiple_d_xdl_cshuffle.hpp" #include "ck/tensor_operation/gpu/device/impl/device_splitk_contraction_multiple_d_xdl_cshuffle.hpp"
#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp" #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
#include "ck/library/utility/check_err.hpp" #include "ck/library/utility/check_err.hpp"
......
example/44_conv2d_fwd_quantization/CMakeLists.txt deleted 100644 → 0
View file @ 9b4c780a
add_example_executable(example_conv2d_fwd_xdl_perchannel_quantization_int8 conv2d_fwd_xdl_bias_relu_perchannel_quantization_int8.cpp)
add_example_executable(example_conv2d_fwd_xdl_perlayer_quantization_int8 conv2d_fwd_xdl_perlayer_quantization_int8.cpp)
add_example_executable(example_conv2d_fwd_xdl_bias_relu_perlayer_quantization_int8 conv2d_fwd_xdl_bias_relu_perlayer_quantization_int8.cpp)
example/44_elementwise_permute/elementwise_permute_4D_fp16.cpp
View file @ 76f2b6cd
...@@ -3,7 +3,7 @@ ...@@ -3,7 +3,7 @@
#include "ck/ck.hpp" #include "ck/ck.hpp"
#include "ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp" #include "ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp"
#include "ck/tensor_operation/gpu/device/impl/device_elementwise.hpp" #include "ck/tensor_operation/gpu/device/impl/device_elementwise_impl.hpp"
#include "ck/library/utility/algorithm.hpp" #include "ck/library/utility/algorithm.hpp"
#include "ck/library/utility/check_err.hpp" #include "ck/library/utility/check_err.hpp"
...@@ -19,13 +19,13 @@ using BDataType = F16; ...@@ -19,13 +19,13 @@ using BDataType = F16;
using PassThrough = ck::tensor_operation::element_wise::PassThrough; using PassThrough = ck::tensor_operation::element_wise::PassThrough;
using DeviceElementwisePermuteInstance = using DeviceElementwisePermuteInstance =
ck::tensor_operation::device::DeviceElementwise<ck::Tuple<ADataType>, ck::tensor_operation::device::DeviceElementwiseImpl<ck::Tuple<ADataType>,
ck::Tuple<BDataType>, ck::Tuple<BDataType>,
PassThrough, PassThrough,
4, 4,
8, 8,
ck::Sequence<8>, ck::Sequence<8>,
ck::Sequence<1>>; ck::Sequence<1>>;
template <typename HostTensorA, typename HostTensorB, typename Functor> template <typename HostTensorA, typename HostTensorB, typename Functor>
void host_elementwise4D(HostTensorB& B_nhwc, const HostTensorA& A_nchw, Functor functor) void host_elementwise4D(HostTensorB& B_nhwc, const HostTensorA& A_nchw, Functor functor)
......
example/44_elementwise_permute/elementwise_permute_4D_fp16_2d.cpp
View file @ 76f2b6cd
...@@ -3,7 +3,7 @@ ...@@ -3,7 +3,7 @@
#include "ck/ck.hpp" #include "ck/ck.hpp"
#include "ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp" #include "ck/tensor_operation/gpu/element/binary_element_wise_operation.hpp"
#include "ck/tensor_operation/gpu/device/device_elementwise_2d.hpp" #include "ck/tensor_operation/gpu/device/impl/device_elementwise_2d_impl.hpp"
#include "ck/library/utility/check_err.hpp" #include "ck/library/utility/check_err.hpp"
#include "ck/library/utility/device_memory.hpp" #include "ck/library/utility/device_memory.hpp"
...@@ -17,15 +17,15 @@ using BDataType = F16; ...@@ -17,15 +17,15 @@ using BDataType = F16;
using PassThrough = ck::tensor_operation::element_wise::PassThrough; using PassThrough = ck::tensor_operation::element_wise::PassThrough;
using DeviceElementwisePermuteInstance = using DeviceElementwisePermuteInstance =
ck::tensor_operation::device::DeviceElementwise<ck::Tuple<ADataType>, ck::tensor_operation::device::DeviceElementwise2dImpl<ck::Tuple<ADataType>,
ck::Tuple<BDataType>, ck::Tuple<BDataType>,
PassThrough, PassThrough,
3, // NumDim_M 3, // NumDim_M
1, // NumDim_N 1, // NumDim_N
8, 8,
8, 8,
ck::Sequence<8>, ck::Sequence<8>,
ck::Sequence<8>>; ck::Sequence<8>>;
template <typename HostTensorA, typename HostTensorB, typename Functor> template <typename HostTensorA, typename HostTensorB, typename Functor>
void host_elementwise4D(HostTensorB& B_nhwc, void host_elementwise4D(HostTensorB& B_nhwc,
......
example/45_elementwise_normalization/elementwise_layernorm_blockwise.cpp
View file @ 76f2b6cd
// SPDX-License-Identifier: MIT // SPDX-License-Identifier: MIT
// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved. // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
#include <iostream> #include <iostream>
#include <numeric> #include <numeric>
......
example/46_gemm_add_multiply/CMakeLists.txt 0 → 100644
View file @ 76f2b6cd
add_example_executable(example_gemm_add_multiply_dl_fp16 gemm_add_multiply_dl_fp16.cpp)
add_example_executable(example_gemm_add_multiply_xdl_fp16 gemm_add_multiply_xdl_fp16.cpp)
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment