Skip to content

GitLab

Commit d5d3788f authored by Jianfeng yan's avatar Jianfeng yan
Browse files

minor changes

parent 43c22b57
Hide whitespace changes
Inline Side-by-side
Showing with 32 additions and 381 deletions
composable_kernel/include/tensor_operation/gridwise_gemm_xdlops_v2r3.hpp
View file @ d5d3788f
...@@ -8,6 +8,7 @@ ...@@ -8,6 +8,7 @@
#include "blockwise_gemm_xdlops.hpp" #include "blockwise_gemm_xdlops.hpp"
#include "blockwise_tensor_slice_transfer_v4r1.hpp" #include "blockwise_tensor_slice_transfer_v4r1.hpp"
#include "threadwise_tensor_slice_transfer_sfcurve.hpp" #include "threadwise_tensor_slice_transfer_sfcurve.hpp"
// #include "threadwise_tensor_slice_transfer.hpp"
#include "gridwise_gemm_pipeline_v1.hpp" #include "gridwise_gemm_pipeline_v1.hpp"
namespace ck { namespace ck {
...@@ -563,29 +564,28 @@ struct GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v2r3 ...@@ -563,29 +564,28 @@ struct GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v2r3
n_thread_data_on_grid_to_n0_n1_n2_adaptor.CalculateBottomIndex( n_thread_data_on_grid_to_n0_n1_n2_adaptor.CalculateBottomIndex(
make_multi_index(n_thread_data_on_grid)); make_multi_index(n_thread_data_on_grid));
auto c_thread_copy = auto c_thread_copy = ThreadwiseTensorSliceTransfer_v1r3_sfcurve<
ThreadwiseTensorSliceTransfer_v1r3_using_space_filling_curve<FloatAcc, FloatAcc,
FloatC, FloatC,
decltype(c_thread_desc_m0_n0_m1_n1_m2_m3_m4_n2), decltype(c_thread_desc_m0_n0_m1_n1_m2_m3_m4_n2),
decltype(c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2), decltype(c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2),
CElementwiseOperation, CElementwiseOperation,
Sequence<M0, N0, I1, I1, M2, I1, M4, I1>, Sequence<M0, N0, I1, I1, M2, I1, M4, I1>,
CThreadTransferSrcDstAccessOrder, CThreadTransferSrcDstAccessOrder,
CThreadTransferSrcDstVectorDim, CThreadTransferSrcDstVectorDim,
CThreadTransferDstScalarPerVector, CThreadTransferDstScalarPerVector,
CGlobalMemoryDataOperation, CGlobalMemoryDataOperation,
1, 1,
true>{ true>{c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2,
c_grid_desc_m0_n0_m1_n1_m2_m3_m4_n2, make_multi_index(m_thread_data_on_grid_idx[I0],
make_multi_index(m_thread_data_on_grid_idx[I0], n_thread_data_on_grid_idx[I0],
n_thread_data_on_grid_idx[I0], m_thread_data_on_grid_idx[I1],
m_thread_data_on_grid_idx[I1], n_thread_data_on_grid_idx[I1],
n_thread_data_on_grid_idx[I1], m_thread_data_on_grid_idx[I2],
m_thread_data_on_grid_idx[I2], m_thread_data_on_grid_idx[I3],
m_thread_data_on_grid_idx[I3], m_thread_data_on_grid_idx[I4],
m_thread_data_on_grid_idx[I4], n_thread_data_on_grid_idx[I2]),
n_thread_data_on_grid_idx[I2]), c_element_op};
c_element_op};
c_thread_copy.Run(c_thread_desc_m0_n0_m1_n1_m2_m3_m4_n2, c_thread_copy.Run(c_thread_desc_m0_n0_m1_n1_m2_m3_m4_n2,
make_tuple(I0, I0, I0, I0, I0, I0, I0, I0), make_tuple(I0, I0, I0, I0, I0, I0, I0, I0),
......
composable_kernel/include/tensor_operation/threadwise_tensor_slice_transfer_sfcurve.hpp
View file @ d5d3788f
...@@ -30,7 +30,7 @@ template <typename SrcData, ...@@ -30,7 +30,7 @@ template <typename SrcData,
index_t DstScalarStrideInVector, index_t DstScalarStrideInVector,
bool DstResetCoordinateAfterRun, bool DstResetCoordinateAfterRun,
typename enable_if<SrcDesc::IsKnownAtCompileTime(), bool>::type = false> typename enable_if<SrcDesc::IsKnownAtCompileTime(), bool>::type = false>
struct ThreadwiseTensorSliceTransfer_v1r3_using_space_filling_curve struct ThreadwiseTensorSliceTransfer_v1r3_sfcurve
{ {
static constexpr index_t nDim = SliceLengths::Size(); static constexpr index_t nDim = SliceLengths::Size();
...@@ -38,7 +38,7 @@ struct ThreadwiseTensorSliceTransfer_v1r3_using_space_filling_curve ...@@ -38,7 +38,7 @@ struct ThreadwiseTensorSliceTransfer_v1r3_using_space_filling_curve
using DstCoord = decltype(make_tensor_coordinate(DstDesc{}, Index{})); using DstCoord = decltype(make_tensor_coordinate(DstDesc{}, Index{}));
__device__ constexpr ThreadwiseTensorSliceTransfer_v1r3_using_space_filling_curve( __device__ constexpr ThreadwiseTensorSliceTransfer_v1r3_sfcurve(
const DstDesc& dst_desc, const DstDesc& dst_desc,
const Index& dst_slice_origin_idx, const Index& dst_slice_origin_idx,
const DstElementwiseOperation& dst_element_op) const DstElementwiseOperation& dst_element_op)
...@@ -98,7 +98,6 @@ struct ThreadwiseTensorSliceTransfer_v1r3_using_space_filling_curve ...@@ -98,7 +98,6 @@ struct ThreadwiseTensorSliceTransfer_v1r3_using_space_filling_curve
constexpr auto num_accesses = SpaceFillingCurve::GetNumOfAccess(); constexpr auto num_accesses = SpaceFillingCurve::GetNumOfAccess();
static_for<0, num_accesses, 1>{}([&](auto idx_1d) { static_for<0, num_accesses, 1>{}([&](auto idx_1d) {
constexpr auto idx_md = SpaceFillingCurve::GetIndex(idx_1d); constexpr auto idx_md = SpaceFillingCurve::GetIndex(idx_1d);
// copy data from src_buf into dst_vector // copy data from src_buf into dst_vector
...@@ -200,4 +199,3 @@ struct ThreadwiseTensorSliceTransfer_v1r3_using_space_filling_curve ...@@ -200,4 +199,3 @@ struct ThreadwiseTensorSliceTransfer_v1r3_using_space_filling_curve
} // namespace ck } // namespace ck
#endif #endif
composable_kernel/include/tensor_operation/threadwise_tensor_slice_transfer_v6r1_sfcurve.hpp
View file @ d5d3788f
...@@ -78,8 +78,8 @@ struct ThreadwiseTensorSliceTransfer_v6r1 ...@@ -78,8 +78,8 @@ struct ThreadwiseTensorSliceTransfer_v6r1
detail::lambda_scalar_per_access<VectorDim, ScalarPerVector>{}, Number<nDim>{}); detail::lambda_scalar_per_access<VectorDim, ScalarPerVector>{}, Number<nDim>{});
using SpaceFillingCurve = SpaceFillingCurve<SliceLengths, using SpaceFillingCurve = SpaceFillingCurve<SliceLengths,
DimAccessOrder, DimAccessOrder,
remove_cv_t<decltype(scalar_per_access)>>; remove_cv_t<decltype(scalar_per_access)>>;
// loop over space-filling curve // loop over space-filling curve
constexpr auto num_accesses = SpaceFillingCurve::GetNumOfAccess(); constexpr auto num_accesses = SpaceFillingCurve::GetNumOfAccess();
...@@ -198,4 +198,3 @@ struct ThreadwiseTensorSliceTransfer_v6r1 ...@@ -198,4 +198,3 @@ struct ThreadwiseTensorSliceTransfer_v6r1
} // namespace ck } // namespace ck
#endif #endif
composable_kernel/include/tensor_operation/threadwise_tensor_slice_transfer_v6r2_sfcurve.hpp
View file @ d5d3788f
...@@ -250,4 +250,3 @@ struct ThreadwiseTensorSliceTransfer_v6r2 ...@@ -250,4 +250,3 @@ struct ThreadwiseTensorSliceTransfer_v6r2
} // namespace ck } // namespace ck
#endif #endif
composable_kernel/include/tensor_operation/threadwise_tensor_slice_transfer_v6r3_sfcurve.hpp
View file @ d5d3788f
...@@ -300,4 +300,3 @@ struct ThreadwiseTensorSliceTransfer_v6r3 ...@@ -300,4 +300,3 @@ struct ThreadwiseTensorSliceTransfer_v6r3
} // namespace ck } // namespace ck
#endif #endif
composable_kernel/include/utility/dynamic_buffer.hpp
View file @ d5d3788f
...@@ -123,12 +123,11 @@ struct DynamicBuffer ...@@ -123,12 +123,11 @@ struct DynamicBuffer
else if constexpr(Op == InMemoryDataOperationEnum_t::AtomicAdd) else if constexpr(Op == InMemoryDataOperationEnum_t::AtomicAdd)
{ {
this->template AtomicAdd<X>(i, is_valid_element, x); this->template AtomicAdd<X>(i, is_valid_element, x);
} }
else if constexpr(Op == InMemoryDataOperationEnum_t::Add) else if constexpr(Op == InMemoryDataOperationEnum_t::Add)
{ {
auto tmp = this->template Get<X>(i, is_valid_element); auto tmp = this->template Get<X>(i, is_valid_element);
this->template Set<X>(i, is_valid_element, x+tmp); this->template Set<X>(i, is_valid_element, x + tmp);
// tmp += x; // tmp += x;
// this->template Set<X>(i, is_valid_element, tmp); // this->template Set<X>(i, is_valid_element, tmp);
} }
......
composable_kernel/include/utility/tensor_space_filling_curve.hpp
View file @ d5d3788f
...@@ -42,7 +42,8 @@ struct SpaceFillingCurve ...@@ -42,7 +42,8 @@ struct SpaceFillingCurve
} }
template <index_t AccessIdx1dBegin, index_t AccessIdx1dEnd> template <index_t AccessIdx1dBegin, index_t AccessIdx1dEnd>
static __device__ __host__ constexpr auto GetStepBetween(Number<AccessIdx1dBegin>, Number<AccessIdx1dEnd>) static __device__ __host__ constexpr auto GetStepBetween(Number<AccessIdx1dBegin>,
Number<AccessIdx1dEnd>)
{ {
static_assert(AccessIdx1dBegin >= 0, "1D index should be non-negative"); static_assert(AccessIdx1dBegin >= 0, "1D index should be non-negative");
static_assert(AccessIdx1dBegin < GetNumOfAccess(), "1D index should be larger than 0"); static_assert(AccessIdx1dBegin < GetNumOfAccess(), "1D index should be larger than 0");
......
device_operation/include/conv_utils.hpp
View file @ d5d3788f
...@@ -39,12 +39,12 @@ std::size_t GetFlops(ck::index_t N, ...@@ -39,12 +39,12 @@ std::size_t GetFlops(ck::index_t N,
std::accumulate(std::begin(output_spatial_lengths), std::accumulate(std::begin(output_spatial_lengths),
std::end(output_spatial_lengths), std::end(output_spatial_lengths),
static_cast<std::size_t>(1), static_cast<std::size_t>(1),
std::multiplies<std::size_t>()) * std::multiplies<std::size_t>()) *
C * C *
std::accumulate(std::begin(filter_spatial_lengths), std::accumulate(std::begin(filter_spatial_lengths),
std::end(filter_spatial_lengths), std::end(filter_spatial_lengths),
static_cast<std::size_t>(1), static_cast<std::size_t>(1),
std::multiplies<std::size_t>()); std::multiplies<std::size_t>());
} }
/** /**
......
example/CMakeLists.txt
View file @ d5d3788f
...@@ -48,7 +48,3 @@ target_link_libraries(gemm_xdl_alpha_beta PRIVATE host_tensor) ...@@ -48,7 +48,3 @@ target_link_libraries(gemm_xdl_alpha_beta PRIVATE host_tensor)
target_link_libraries(conv2d_fwd_xdl_int8 PRIVATE host_tensor) target_link_libraries(conv2d_fwd_xdl_int8 PRIVATE host_tensor)
target_link_libraries(conv3d_fwd_xdl PRIVATE host_tensor) target_link_libraries(conv3d_fwd_xdl PRIVATE host_tensor)
target_link_libraries(convnd_fwd_xdl PRIVATE host_tensor) target_link_libraries(convnd_fwd_xdl PRIVATE host_tensor)
set(CONV2D_FWD_XDL_SFCURVE_SOURCE conv2d_fwd_xdl_sfcurve/conv2d_fwd_xdl_sfcurve.cpp)
add_executable(conv2d_fwd_xdl_sfcurve ${CONV2D_FWD_XDL_SFCURVE_SOURCE})
target_link_libraries(conv2d_fwd_xdl_sfcurve PRIVATE host_tensor)
example/conv2d_fwd_xdl_sfcurve/README.md deleted 100644 → 0
View file @ 43c22b57
# Instructions for ```conv2d_fwd_xdl``` Example
## Docker script
```bash
docker run \
-it \
--rm \
--privileged \
--group-add sudo \
-w /root/workspace \
-v ${PATH_TO_LOCAL_WORKSPACE}:/root/workspace \
rocm/tensorflow:rocm4.3.1-tf2.6-dev \
/bin/bash
```
## Build ```conv2d_fwd_xdl```
```bash
mkdir build && cd build
```
```bash
# Need to specify target ID, example below is gfx908
cmake \
-D BUILD_DEV=OFF \
-D CMAKE_BUILD_TYPE=Release \
-D CMAKE_CXX_FLAGS="-DCK_AMD_GPU_GFX908 --amdgpu-target=gfx908 -O3 " \
-D CMAKE_CXX_COMPILER=/opt/rocm/bin/hipcc \
-D CMAKE_PREFIX_PATH=/opt/rocm \
..
```
```bash
make -j conv2d_fwd_xdl
```
## Run ```conv2d_fwd_xdl_int8```
```bash
#arg1: verification (0=no, 1=yes)
#arg2: initialization (0=no init, 1=integer value, 2=decimal value)
#arg3: run kernel # of times (>1)
#arg4 to 18: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, RightPx
./example/conv2d_fwd_xdl_int8 0 1 5
```
Result (MI100 @ 1087Mhz, 133.5TFlops peak FP16)
```
in_n_c_hi_wi: dim 4, lengths {128, 192, 71, 71}, strides {967872, 1, 13632, 192}
wei_k_c_y_x: dim 4, lengths {256, 192, 3, 3}, strides {1728, 1, 576, 192}
out_n_k_ho_wo: dim 4, lengths {128, 256, 36, 36}, strides {331776, 1, 9216, 256}
arg.a_grid_desc_k0_m_k1_{216, 165888, 8}
arg.b_grid_desc_k0_n_k1_{216, 256, 8}
arg.c_grid_desc_m_n_{ 165888, 256}
launch_and_time_kernel: grid_dim {1296, 1, 1}, block_dim {256, 1, 1}
Warm up
Start running 5 times...
Perf: 1.43206 ms, 102.486 TFlops, 232.947 GB/s
```
example/conv2d_fwd_xdl_sfcurve/conv2d_fwd_xdl_sfcurve.cpp deleted 100644 → 0
View file @ 43c22b57
#include <iostream>
#include <numeric>
#include <initializer_list>
#include <cstdlib>
#include <stdlib.h>
#include <half.hpp>
#include "config.hpp"
#include "print.hpp"
#include "device.hpp"
#include "host_tensor.hpp"
#include "host_tensor_generator.hpp"
#include "device_tensor.hpp"
#include "tensor_layout.hpp"
#include "device_conv2d_fwd_xdl_nhwc_kyxc_nhwk.hpp"
#include "element_wise_operation.hpp"
#include "reference_conv_fwd.hpp"
#include "convolution_utility.hpp"
// using InDataType = int8_t;
// using WeiDataType = int8_t;
// using OutDataType = int8_t;
// using AccDataType = int32_t;
using InDataType = ck::half_t;
using WeiDataType = ck::half_t;
using OutDataType = ck::half_t;
using AccDataType = float;
template <ck::index_t... Is>
using S = ck::Sequence<Is...>;
using InLayout = ck::tensor_layout::convolution::NHWC;
using WeiLayout = ck::tensor_layout::convolution::KYXC;
using OutLayout = ck::tensor_layout::convolution::NHWK;
using InElementOp = ck::tensor_operation::element_wise::PassThrough;
using WeiElementOp = ck::tensor_operation::element_wise::PassThrough;
using OutElementOp = ck::tensor_operation::element_wise::PassThrough;
using PassThrough = ck::tensor_operation::element_wise::PassThrough;
static constexpr auto ConvFwdDefault =
ck::tensor_operation::device::ConvolutionForwardSpecialization_t::Default;
using DeviceConvFwdInstance = ck::tensor_operation::device::
DeviceConv2dFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<
InDataType, // InDataType
WeiDataType, // WeiDataType
OutDataType, // OutDataType
AccDataType, // AccDataType
PassThrough, // InElementwiseOperation
PassThrough, // WeiElementwiseOperation
PassThrough, // OutElementwiseOperation
ConvFwdDefault, // ConvForwardSpecialization
256, // BlockSize
128, // MPerBlock
256, // NPerBlock
4, // K0PerBlock
8, // K1
32, // MPerXdl
32, // NPerXdl
2, // MXdlPerWave
4, // NXdlPerWave
S<4, 64, 1>, // ABlockTransferThreadClusterLengths_K0_M_K1
S<1, 0, 2>, // ABlockTransferThreadClusterArrangeOrder
S<1, 0, 2>, // ABlockTransferSrcAccessOrder
2, // ABlockTransferSrcVectorDim
8, // ABlockTransferSrcScalarPerVector
8, // ABlockTransferDstScalarPerVector_K1
true, // ABlockLdsAddExtraM
S<4, 64, 1>, // BBlockTransferThreadClusterLengths_K0_N_K1
S<1, 0, 2>, // BBlockTransferThreadClusterArrangeOrder
S<1, 0, 2>, // BBlockTransferSrcAccessOrder
2, // BBlockTransferSrcVectorDim
8, // BBlockTransferSrcScalarPerVector
8, // BBlockTransferDstScalarPerVector_K1
true, // BBlockLdsAddExtraN
7, // CThreadTransferSrcDstVectorDim
1>; // CThreadTransferDstScalarPerVector
using ReferenceConvFwdInstance = ck::tensor_operation::host::
ReferenceConvFwd<InDataType, WeiDataType, OutDataType, InElementOp, WeiElementOp, OutElementOp>;
int main(int argc, char* argv[])
{
bool do_verification = 0;
int init_method = 0;
int nrepeat = 5;
// Conv shape
ck::index_t N = 128;
ck::index_t K = 256;
ck::index_t C = 192;
ck::index_t Y = 3;
ck::index_t X = 3;
ck::index_t Hi = 71;
ck::index_t Wi = 71;
ck::index_t conv_stride_h = 2;
ck::index_t conv_stride_w = 2;
ck::index_t conv_dilation_h = 1;
ck::index_t conv_dilation_w = 1;
ck::index_t in_left_pad_h = 1;
ck::index_t in_left_pad_w = 1;
ck::index_t in_right_pad_h = 1;
ck::index_t in_right_pad_w = 1;
if(argc == 4)
{
do_verification = std::stoi(argv[1]);
init_method = std::stoi(argv[2]);
nrepeat = std::stoi(argv[3]);
}
else if(argc == 19)
{
do_verification = std::stoi(argv[1]);
init_method = std::stoi(argv[2]);
nrepeat = std::stoi(argv[3]);
N = std::stoi(argv[4]);
K = std::stoi(argv[5]);
C = std::stoi(argv[6]);
Y = std::stoi(argv[7]);
X = std::stoi(argv[8]);
Hi = std::stoi(argv[9]);
Wi = std::stoi(argv[10]);
conv_stride_h = std::stoi(argv[11]);
conv_stride_w = std::stoi(argv[12]);
conv_dilation_h = std::stoi(argv[13]);
conv_dilation_w = std::stoi(argv[14]);
in_left_pad_h = std::stoi(argv[15]);
in_left_pad_w = std::stoi(argv[16]);
in_right_pad_h = std::stoi(argv[17]);
in_right_pad_w = std::stoi(argv[18]);
}
else
{
printf("arg1: verification (0=no, 1=yes)\n");
printf("arg2: initialization (0=no init, 1=integer value, 2=decimal value)\n");
printf("arg3: run kernel # of times (>1)\n");
printf("arg4 to 18: N, K, C, Y, X, Hi, Wi, Sy, Sx, Dy, Dx, LeftPy, LeftPx, RightPy, "
"RightPx\n");
exit(0);
}
const std::vector<ck::index_t> conv_filter_strides{conv_stride_h, conv_stride_w};
const std::vector<ck::index_t> conv_filter_dilations{conv_dilation_h, conv_dilation_w};
const std::vector<ck::index_t> input_left_pads{in_left_pad_h, in_left_pad_w};
const std::vector<ck::index_t> input_right_pads{in_right_pad_h, in_right_pad_w};
const auto output_spatial_lengths =
ck::tensor_operation::ConvolutionUtility::ComputeOutputSpatialLengths({Hi, Wi},
{Y, X},
conv_filter_strides,
conv_filter_dilations,
input_left_pads,
input_right_pads);
const ck::index_t Ho = output_spatial_lengths[0];
const ck::index_t Wo = output_spatial_lengths[1];
// tensor layout
auto f_host_tensor_descriptor = [](std::size_t N_,
std::size_t C_,
std::size_t H,
std::size_t W,
auto layout) {
if constexpr(ck::is_same<decltype(layout), ck::tensor_layout::convolution::NCHW>::value ||
ck::is_same<decltype(layout), ck::tensor_layout::convolution::KCYX>::value ||
ck::is_same<decltype(layout), ck::tensor_layout::convolution::NKHW>::value)
{
return HostTensorDescriptor(std::vector<std::size_t>({N_, C_, H, W}),
std::vector<std::size_t>({C_ * H * W, H * W, W, 1}));
}
else if constexpr(ck::is_same<decltype(layout),
ck::tensor_layout::convolution::NHWC>::value ||
ck::is_same<decltype(layout),
ck::tensor_layout::convolution::KYXC>::value ||
ck::is_same<decltype(layout),
ck::tensor_layout::convolution::NHWK>::value)
{
return HostTensorDescriptor(std::vector<std::size_t>({N_, C_, H, W}),
std::vector<std::size_t>({C_ * H * W, 1, W * C_, C_}));
}
};
Tensor<InDataType> in_n_c_hi_wi(f_host_tensor_descriptor(N, C, Hi, Wi, InLayout{}));
Tensor<WeiDataType> wei_k_c_y_x(f_host_tensor_descriptor(K, C, Y, X, WeiLayout{}));
Tensor<OutDataType> out_n_k_ho_wo_host_result(
f_host_tensor_descriptor(N, K, Ho, Wo, OutLayout{}));
Tensor<OutDataType> out_n_k_ho_wo_device_result(
f_host_tensor_descriptor(N, K, Ho, Wo, OutLayout{}));
std::cout << "in_n_c_hi_wi: " << in_n_c_hi_wi.mDesc << std::endl;
std::cout << "wei_k_c_y_x: " << wei_k_c_y_x.mDesc << std::endl;
std::cout << "out_n_k_ho_wo: " << out_n_k_ho_wo_host_result.mDesc << std::endl;
switch(init_method)
{
case 0: break;
case 1:
in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_2<InDataType>{-1, 1});
wei_k_c_y_x.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-1, 1});
break;
default:
in_n_c_hi_wi.GenerateTensorValue(GeneratorTensor_3<InDataType>{0, 1});
wei_k_c_y_x.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-1, 1});
}
DeviceMem in_device_buf(sizeof(InDataType) * in_n_c_hi_wi.mDesc.GetElementSpace());
DeviceMem wei_device_buf(sizeof(WeiDataType) * wei_k_c_y_x.mDesc.GetElementSpace());
DeviceMem out_device_buf(sizeof(OutDataType) *
out_n_k_ho_wo_device_result.mDesc.GetElementSpace());
in_device_buf.ToDevice(in_n_c_hi_wi.mData.data());
wei_device_buf.ToDevice(wei_k_c_y_x.mData.data());
{
memset(out_n_k_ho_wo_device_result.mData.data(), static_cast<OutDataType>(0), out_n_k_ho_wo_device_result.mDesc.GetElementSpace());
out_device_buf.ToDevice(out_n_k_ho_wo_device_result.mData.data());
}
// do GEMM
auto conv = DeviceConvFwdInstance{};
auto invoker = conv.MakeInvoker();
auto argument = conv.MakeArgument(static_cast<InDataType*>(in_device_buf.GetDeviceBuffer()),
static_cast<WeiDataType*>(wei_device_buf.GetDeviceBuffer()),
static_cast<OutDataType*>(out_device_buf.GetDeviceBuffer()),
N,
K,
C,
std::vector<ck::index_t>{Hi, Wi},
std::vector<ck::index_t>{Y, X},
std::vector<ck::index_t>{Ho, Wo},
conv_filter_strides,
conv_filter_dilations,
input_left_pads,
input_right_pads,
InElementOp{},
WeiElementOp{},
OutElementOp{});
if(!conv.IsSupportedArgument(argument))
{
throw std::runtime_error(
"wrong! device_conv with the specified compilation parameters does "
"not support this Conv problem");
}
float ave_time = invoker.Run(argument, nrepeat);
std::size_t flop = std::size_t(2) * N * K * Ho * Wo * C * Y * X;
std::size_t num_btype = sizeof(InDataType) * (N * C * Hi * Wi) +
sizeof(WeiDataType) * (K * C * Y * X) +
sizeof(OutDataType) * (N * K * Ho * Wo);
float tflops = static_cast<float>(flop) / 1.E9 / ave_time;
float gb_per_sec = num_btype / 1.E6 / ave_time;
std::cout << "Perf: " << ave_time << " ms, " << tflops << " TFlops, " << gb_per_sec << " GB/s"
<< std::endl;
if(do_verification)
{
auto ref_conv = ReferenceConvFwdInstance{};
auto ref_invoker = ref_conv.MakeInvoker();
auto ref_argument = ref_conv.MakeArgument(in_n_c_hi_wi,
wei_k_c_y_x,
out_n_k_ho_wo_host_result,
conv_filter_strides,
conv_filter_dilations,
input_left_pads,
input_right_pads,
InElementOp{},
WeiElementOp{},
OutElementOp{});
ref_invoker.Run(ref_argument);
out_device_buf.FromDevice(out_n_k_ho_wo_device_result.mData.data());
check_error(out_n_k_ho_wo_host_result, out_n_k_ho_wo_device_result);
}
}
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