Support erf and erfc ant GPU backend

This commit is contained in:
peixu_ren 2020-09-09 20:51:43 -04:00
parent 3fe8916afa
commit fdd2d8209f
10 changed files with 438 additions and 0 deletions

View File

@ -0,0 +1,32 @@
/**
* Copyright 2020 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "erf_impl.cuh"
template <typename T>
__global__ void ErfKernel(T *input, T *output, size_t count) {
for (size_t i = blockIdx.x * blockDim.x + threadIdx.x; i < (count); i += blockDim.x * gridDim.x) {
output[i] = (T)erf(input[i]);
}
return;
}
template <typename T>
void Erf(T *input, T *output, size_t count, cudaStream_t cuda_stream) {
ErfKernel<<<GET_BLOCKS(count), GET_THREADS, 0, cuda_stream>>>(input, output, count);
return;
}
template void Erf<float>(float *input, float *output, size_t count, cudaStream_t cuda_stream);

View File

@ -0,0 +1,25 @@
/**
* Copyright 2020 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#ifndef MINDSPORE_CCSRC_KERNEL_GPU_CUDA_IMPL_ERFIMPL_H_
#define MINDSPORE_CCSRC_KERNEL_GPU_CUDA_IMPL_ERFIMPL_H_
#include <curand_kernel.h>
#include "runtime/device/gpu/cuda_common.h"
template <typename T>
void Erf(T *input, T *output, size_t count, cudaStream_t cuda_stream);
#endif // MINDSPORE_CCSRC_KERNEL_GPU_CUDA_IMPL_ERFIMPL_H_

View File

@ -0,0 +1,32 @@
/**
* Copyright 2020 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "erfc_impl.cuh"
template <typename T>
__global__ void ErfcKernel(T *input, T *output, size_t count) {
for (size_t i = blockIdx.x * blockDim.x + threadIdx.x; i < (count); i += blockDim.x * gridDim.x) {
output[i] = (T)erfc(input[i]);
}
return;
}
template <typename T>
void Erfc(T *input, T *output, size_t count, cudaStream_t cuda_stream) {
ErfcKernel<<<GET_BLOCKS(count), GET_THREADS, 0, cuda_stream>>>(input, output, count);
return;
}
template void Erfc<float>(float *input, float *output, size_t count, cudaStream_t cuda_stream);

View File

@ -0,0 +1,25 @@
/**
* Copyright 2020 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#ifndef MINDSPORE_CCSRC_KERNEL_GPU_CUDA_IMPL_ERFIMPL_H_
#define MINDSPORE_CCSRC_KERNEL_GPU_CUDA_IMPL_ERFIMPL_H_
#include <curand_kernel.h>
#include "runtime/device/gpu/cuda_common.h"
template <typename T>
void Erfc(T *input, T *output, size_t count, cudaStream_t cuda_stream);
#endif // MINDSPORE_CCSRC_KERNEL_GPU_CUDA_IMPL_ERFIMPL_H_

View File

@ -0,0 +1,24 @@
/**
* Copyright 2020 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "backend/kernel_compiler/gpu/math/erf_gpu_kernel.h"
namespace mindspore {
namespace kernel {
MS_REG_GPU_KERNEL_ONE(Erf, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeFloat32),
ErfGpuKernel, float)
} // namespace kernel
} // namespace mindspore

View File

@ -0,0 +1,92 @@
/**
* Copyright 2020 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#ifndef MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_GPU_ERF_GPU_KERNEL_H_
#define MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_GPU_ERF_GPU_KERNEL_H_
#include <cuda_runtime_api.h>
#include <vector>
#include <string>
#include "backend/kernel_compiler/gpu/gpu_kernel.h"
#include "backend/kernel_compiler/gpu/gpu_kernel_factory.h"
#include "backend/kernel_compiler/gpu/cuda_impl/erf_impl.cuh"
namespace mindspore {
namespace kernel {
template <typename T>
class ErfGpuKernel : public GpuKernel {
public:
ErfGpuKernel() : input_size_(sizeof(T)), output_size_(sizeof(T)) {}
~ErfGpuKernel() override = default;
const std::vector<size_t> &GetInputSizeList() const override { return input_size_list_; }
const std::vector<size_t> &GetOutputSizeList() const override { return output_size_list_; }
const std::vector<size_t> &GetWorkspaceSizeList() const override { return workspace_size_list_; }
bool Launch(const std::vector<AddressPtr> &inputs, const std::vector<AddressPtr> &workspace,
const std::vector<AddressPtr> &outputs, void *stream_ptr) override {
VARIABLE_NOT_USED(workspace);
T *input_addr = GetDeviceAddress<T>(inputs, 0);
T *output_addr = GetDeviceAddress<T>(outputs, 0);
Erf(input_addr, output_addr, outputs[0]->size / sizeof(T), reinterpret_cast<cudaStream_t>(stream_ptr));
return true;
}
bool Init(const CNodePtr &kernel_node) override {
size_t input_num = AnfAlgo::GetInputTensorNum(kernel_node);
if (input_num != 1) {
MS_LOG(ERROR) << "Input number is " << input_num << ", but erf needs 3 inputs.";
return false;
}
size_t output_num = AnfAlgo::GetOutputTensorNum(kernel_node);
if (output_num != 1) {
MS_LOG(ERROR) << "Output number is " << output_num << ", but erf needs 1 output.";
return false;
}
auto input_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 0);
for (size_t i = 0; i < input_shape.size(); i++) {
input_size_ *= input_shape[i];
}
auto output_shape = AnfAlgo::GetOutputInferShape(kernel_node, 0);
for (size_t i = 0; i < output_shape.size(); i++) {
output_size_ *= output_shape[i];
}
if (input_size_ != output_size_) {
MS_LOG(ERROR) << "Input size and output should be equal for Erf.";
return false;
}
InitSizeLists();
return true;
}
protected:
void InitSizeLists() override {
input_size_list_.push_back(input_size_);
output_size_list_.push_back(output_size_);
}
private:
size_t input_size_;
size_t output_size_;
std::vector<size_t> input_size_list_;
std::vector<size_t> output_size_list_;
std::vector<size_t> workspace_size_list_;
};
} // namespace kernel
} // namespace mindspore
#endif // MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_GPU_ERF_GPU_KERNEL_H_

View File

@ -0,0 +1,24 @@
/**
* Copyright 2020 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "backend/kernel_compiler/gpu/math/erfc_gpu_kernel.h"
namespace mindspore {
namespace kernel {
MS_REG_GPU_KERNEL_ONE(Erfc, KernelAttr().AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeFloat32),
ErfcGpuKernel, float)
} // namespace kernel
} // namespace mindspore

View File

@ -0,0 +1,92 @@
/**
* Copyright 2020 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#ifndef MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_GPU_ERF_GPU_KERNEL_H_
#define MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_GPU_ERF_GPU_KERNEL_H_
#include <cuda_runtime_api.h>
#include <vector>
#include <string>
#include "backend/kernel_compiler/gpu/gpu_kernel.h"
#include "backend/kernel_compiler/gpu/gpu_kernel_factory.h"
#include "backend/kernel_compiler/gpu/cuda_impl/erfc_impl.cuh"
namespace mindspore {
namespace kernel {
template <typename T>
class ErfcGpuKernel : public GpuKernel {
public:
ErfcGpuKernel() : input_size_(sizeof(T)), output_size_(sizeof(T)) {}
~ErfcGpuKernel() override = default;
const std::vector<size_t> &GetInputSizeList() const override { return input_size_list_; }
const std::vector<size_t> &GetOutputSizeList() const override { return output_size_list_; }
const std::vector<size_t> &GetWorkspaceSizeList() const override { return workspace_size_list_; }
bool Launch(const std::vector<AddressPtr> &inputs, const std::vector<AddressPtr> &workspace,
const std::vector<AddressPtr> &outputs, void *stream_ptr) override {
VARIABLE_NOT_USED(workspace);
T *input_addr = GetDeviceAddress<T>(inputs, 0);
T *output_addr = GetDeviceAddress<T>(outputs, 0);
Erfc(input_addr, output_addr, outputs[0]->size / sizeof(T), reinterpret_cast<cudaStream_t>(stream_ptr));
return true;
}
bool Init(const CNodePtr &kernel_node) override {
size_t input_num = AnfAlgo::GetInputTensorNum(kernel_node);
if (input_num != 1) {
MS_LOG(ERROR) << "Input number is " << input_num << ", but erfc needs 3 inputs.";
return false;
}
size_t output_num = AnfAlgo::GetOutputTensorNum(kernel_node);
if (output_num != 1) {
MS_LOG(ERROR) << "Output number is " << output_num << ", but erfc needs 1 output.";
return false;
}
auto input_shape = AnfAlgo::GetPrevNodeOutputInferShape(kernel_node, 0);
for (size_t i = 0; i < input_shape.size(); i++) {
input_size_ *= input_shape[i];
}
auto output_shape = AnfAlgo::GetOutputInferShape(kernel_node, 0);
for (size_t i = 0; i < output_shape.size(); i++) {
output_size_ *= output_shape[i];
}
if (input_size_ != output_size_) {
MS_LOG(ERROR) << "Input size and output should be equal for Erfc.";
return false;
}
InitSizeLists();
return true;
}
protected:
void InitSizeLists() override {
input_size_list_.push_back(input_size_);
output_size_list_.push_back(output_size_);
}
private:
size_t input_size_;
size_t output_size_;
std::vector<size_t> input_size_list_;
std::vector<size_t> output_size_list_;
std::vector<size_t> workspace_size_list_;
};
} // namespace kernel
} // namespace mindspore
#endif // MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_GPU_ERF_GPU_KERNEL_H_

View File

@ -0,0 +1,46 @@
# Copyright 2020 Huawei Technologies Co., Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================
import numpy as np
import pytest
from scipy import special
import mindspore.context as context
import mindspore.nn as nn
from mindspore import Tensor
from mindspore.ops import operations as P
from mindspore import dtype
context.set_context(mode=context.GRAPH_MODE, device_target="GPU")
class NetErf(nn.Cell):
def __init__(self):
super(NetErf, self).__init__()
self.erf = P.Erf()
def construct(self, x):
return self.erf(x)
@pytest.mark.level0
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard
def test_exp():
erf = NetErf()
x = np.array([2.0, 3.0, 4.0, 5.0]).astype(np.float32)
output = erf(Tensor(x, dtype=dtype.float32))
expect = special.erf(x)
tol = 1e-6
assert (np.abs(output.asnumpy() - expect) < tol).all()

View File

@ -0,0 +1,46 @@
# Copyright 2020 Huawei Technologies Co., Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================
import numpy as np
import pytest
from scipy import special
import mindspore.context as context
import mindspore.nn as nn
from mindspore import Tensor
from mindspore.ops import operations as P
from mindspore import dtype
context.set_context(mode=context.GRAPH_MODE, device_target="GPU")
class NetErfc(nn.Cell):
def __init__(self):
super(NetErfc, self).__init__()
self.erfc = P.Erfc()
def construct(self, x):
return self.erfc(x)
@pytest.mark.level0
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard
def test_exp():
erfc = NetErfc()
x = np.array([2.0, 3.0, 4.0, 5.0]).astype(np.float32)
output = erfc(Tensor(x, dtype=dtype.float32))
expect = special.erfc(x)
tol = 1e-6
assert (np.abs(output.asnumpy() - expect) < tol).all()