!21131 sponge ops 0730

Merge pull request !21131 from jiahongQian/master1

mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/sponge/common_sponge.cuh

@@ -359,4 +359,8 @@ __global__ static void Print(const size_t size, const int *input_x) {
   return;
 }
 
+__device__ static VECTOR Make_Vector_Not_Exceed_Value(VECTOR vector, const float value) {
+  return fminf(1.0, value * rnorm3df(vector.x, vector.y, vector.z)) * vector;
+}
+
 #endif  // MINDSPORE_CCSRC_KERNEL_GPU_CUDA_IMPL_SPONGE_COMMON_SPONGE_H_

mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/sponge/nvtit/md_iteration_gradient_descent_impl.cu

@@ -0,0 +1,41 @@
+/**
+ * Copyright 2021 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/cuda_impl/sponge/nvtit/md_iteration_gradient_descent_impl.cuh"
+#include "backend/kernel_compiler/gpu/cuda_impl/util.cuh"
+#include "backend/kernel_compiler/gpu/cuda_impl/sponge/common_sponge.cuh"
+
+__global__ void MD_Iteration_Gradient_Descent(const int atom_numbers, VECTOR *crd, VECTOR *frc,
+                                              const float learning_rate) {
+  int i = blockDim.x * blockIdx.x + threadIdx.x;
+  if (i < atom_numbers) {
+    crd[i].x = crd[i].x + learning_rate * frc[i].x;
+    crd[i].y = crd[i].y + learning_rate * frc[i].y;
+    crd[i].z = crd[i].z + learning_rate * frc[i].z;
+
+    frc[i].x = 0.;
+    frc[i].y = 0.;
+    frc[i].z = 0.;
+  }
+}
+
+void MDIterationGradientDescent(const int atom_numbers, float *crd, float *frc, const float learning_rate,
+                                cudaStream_t stream) {
+  VECTOR *d_crd = reinterpret_cast<VECTOR *>(crd);
+  VECTOR *d_frc = reinterpret_cast<VECTOR *>(frc);
+  MD_Iteration_Gradient_Descent<<<ceilf(static_cast<float>(atom_numbers) / 128), 128, 0, stream>>>(
+    atom_numbers, d_crd, d_frc, learning_rate);
+}

mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/sponge/nvtit/md_iteration_gradient_descent_impl.cuh

@@ -0,0 +1,25 @@
+/**
+ * Copyright 2021 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_SPONGE_MD_ITERATION_GRADIENT_DESCENT_IMPL_H_
+#define MINDSPORE_CCSRC_KERNEL_GPU_CUDA_IMPL_SPONGE_MD_ITERATION_GRADIENT_DESCENT_IMPL_H_
+
+#include <curand_kernel.h>
+#include "runtime/device/gpu/cuda_common.h"
+void MDIterationGradientDescent(const int atom_numbers, float *crd, float *frc, const float learning_rate,
+                                cudaStream_t stream);
+
+#endif

mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/sponge/nvtit/md_iteration_leap_frog_liujian_gpu_impl.cu

@@ -54,7 +54,7 @@ void MD_Iteration_Leap_Frog_With_LiuJian(const int atom_numbers, const float hal
                                          curandStatePhilox4_32_10_t *rand_state, float *rand_frc, float *output,
                                          cudaStream_t stream) {
   Rand_Normal<<<ceilf(static_cast<float>(float4_numbers) / 32.), 32, 0, stream>>>(float4_numbers, rand_state,
-                                                                                 reinterpret_cast<float4 *>(rand_frc));
+                                                                                  reinterpret_cast<float4 *>(rand_frc));
   VECTOR *d_vel = reinterpret_cast<VECTOR *>(vel);
   VECTOR *d_crd = reinterpret_cast<VECTOR *>(crd);
   VECTOR *d_frc = reinterpret_cast<VECTOR *>(frc);

mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/sponge/nvtit/md_iteration_leap_frog_with_max_vel_impl.cu

@@ -0,0 +1,44 @@
+/**
+ * Copyright 2021 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/cuda_impl/sponge/nvtit/md_iteration_leap_frog_with_max_vel_impl.cuh"
+#include "backend/kernel_compiler/gpu/cuda_impl/util.cuh"
+#include "backend/kernel_compiler/gpu/cuda_impl/sponge/common_sponge.cuh"
+
+__global__ void MD_Iteration_Leap_Frog_With_Max_Velocity(const int atom_numbers, VECTOR *vel, VECTOR *crd, VECTOR *frc,
+                                                         VECTOR *acc, const float *inverse_mass, const float dt,
+                                                         const float max_velocity) {
+  int i = blockDim.x * blockIdx.x + threadIdx.x;
+  if (i < atom_numbers) {
+    VECTOR acc_i = inverse_mass[i] * frc[i];
+    VECTOR vel_i = vel[i] + dt * acc_i;
+    vel_i = Make_Vector_Not_Exceed_Value(vel_i, max_velocity);
+    vel[i] = vel_i;
+    crd[i] = crd[i] + dt * vel_i;
+    frc[i] = {0.0f, 0.0f, 0.0f};
+  }
+}
+
+void MDIterationLeapFrogWithMaxVelocity(const int atom_numbers, float *vel, float *crd, float *frc, float *acc,
+                                        const float *inverse_mass, const float dt, const float max_velocity,
+                                        cudaStream_t stream) {
+  VECTOR *d_vel = reinterpret_cast<VECTOR *>(vel);
+  VECTOR *d_crd = reinterpret_cast<VECTOR *>(crd);
+  VECTOR *d_frc = reinterpret_cast<VECTOR *>(frc);
+  VECTOR *d_acc = reinterpret_cast<VECTOR *>(acc);
+  MD_Iteration_Leap_Frog_With_Max_Velocity<<<ceilf(static_cast<float>(atom_numbers) / 128), 128, 0, stream>>>(
+    atom_numbers, d_vel, d_crd, d_frc, d_acc, inverse_mass, dt, max_velocity);
+}

mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/sponge/nvtit/md_iteration_leap_frog_with_max_vel_impl.cuh

@@ -0,0 +1,26 @@
+/**
+ * Copyright 2021 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_SPONGE_MD_ITERATION_LEAP_FROG_WITH_MAX_VEL_IMPL_H_
+#define MINDSPORE_CCSRC_KERNEL_GPU_CUDA_IMPL_SPONGE_MD_ITERATION_LEAP_FROG_WITH_MAX_VEL_IMPL_H_
+
+#include <curand_kernel.h>
+#include "runtime/device/gpu/cuda_common.h"
+void MDIterationLeapFrogWithMaxVelocity(const int atom_numbers, float *vel, float *crd, float *frc, float *acc,
+                                        const float *inverse_mass, const float dt, const float max_velocity,
+                                        cudaStream_t stream);
+
+#endif  // MINDSPORE_CCSRC_KERNEL_GPU_CUDA_IMPL_SPONGE_MD_ITERATION_LEAP_FROG_WITH_MAX_VEL_IMPL_H_

mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/sponge/nvtit/md_iteration_setup_random_state_gpu_impl.cu

@@ -21,7 +21,7 @@
 void MD_Iteration_Setup_Random_State(int float4_numbers, curandStatePhilox4_32_10_t *rand_state, int seed,
                                      cudaStream_t stream) {
   Setup_Rand_Normal_Kernel<<<ceilf(static_cast<float>(float4_numbers) / 32.), 32, 0, stream>>>(float4_numbers,
-                                                                                              rand_state, seed);
+                                                                                               rand_state, seed);
 }
 
 void MD_Iteration_Setup_Random_State(int float4_numbers, curandStatePhilox4_32_10_t *rand_state, int seed,

mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/sponge/pme/pme_common.cuh

@@ -108,16 +108,16 @@ __global__ static void PME_Atom_Near(const UNSIGNED_INT_VECTOR *uint_crd, int *P
     UNSIGNED_INT_VECTOR *temp_uxyz = &PME_uxyz[atom];
     int k, tempux, tempuy, tempuz;
     float tempf;
-    tempf = static_cast<float> (uint_crd[atom].uint_x) * periodic_factor_inverse_x;
+    tempf = static_cast<float>(uint_crd[atom].uint_x) * periodic_factor_inverse_x;
-    tempux = static_cast<int> (tempf);
+    tempux = static_cast<int>(tempf);
     PME_frxyz[atom].x = tempf - tempux;
 
-    tempf = static_cast<float> (uint_crd[atom].uint_y) * periodic_factor_inverse_y;
+    tempf = static_cast<float>(uint_crd[atom].uint_y) * periodic_factor_inverse_y;
-    tempuy = static_cast<int> (tempf);
+    tempuy = static_cast<int>(tempf);
     PME_frxyz[atom].y = tempf - tempuy;
 
-    tempf = static_cast<float> (uint_crd[atom].uint_z) * periodic_factor_inverse_z;
+    tempf = static_cast<float>(uint_crd[atom].uint_z) * periodic_factor_inverse_z;
-    tempuz = static_cast<int> (tempf);
+    tempuz = static_cast<int>(tempf);
     PME_frxyz[atom].z = tempf - tempuz;
 
     if (tempux != (*temp_uxyz).uint_x || tempuy != (*temp_uxyz).uint_y || tempuz != (*temp_uxyz).uint_z) {

mindspore/ccsrc/backend/kernel_compiler/gpu/sponge/nvtit/md_iteration_gradient_descent_kernel.cc

@@ -0,0 +1,26 @@
+/**
+ * Copyright 2021 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/sponge/nvtit/md_iteration_gradient_descent_kernel.h"
+
+namespace mindspore {
+namespace kernel {
+MS_REG_GPU_KERNEL_ONE(
+  MDIterationGradientDescent,
+  KernelAttr().AddInputAttr(kNumberTypeFloat32).AddInputAttr(kNumberTypeFloat32).AddOutputAttr(kNumberTypeFloat32),
+  MDIterationGradientDescentGpuKernel, float)
+}  // namespace kernel
+}  // namespace mindspore

mindspore/ccsrc/backend/kernel_compiler/gpu/sponge/nvtit/md_iteration_gradient_descent_kernel.h

@@ -0,0 +1,77 @@
+/**
+ * Copyright 2021 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_MD_ITERATION_GRADIENT_DESCENT_KERNEL_H_
+#define MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_GPU_MD_ITERATION_GRADIENT_DESCENT_KERNEL_H_
+
+#include "backend/kernel_compiler/gpu/cuda_impl/sponge/nvtit/md_iteration_gradient_descent_impl.cuh"
+#include <cuda_runtime_api.h>
+#include <map>
+#include <string>
+#include <vector>
+
+#include "backend/kernel_compiler/gpu/gpu_kernel.h"
+#include "backend/kernel_compiler/gpu/gpu_kernel_factory.h"
+#include "runtime/device/gpu/cuda_common.h"
+
+namespace mindspore {
+namespace kernel {
+template <typename T>
+class MDIterationGradientDescentGpuKernel : public GpuKernel {
+ public:
+  MDIterationGradientDescentGpuKernel() {}
+  ~MDIterationGradientDescentGpuKernel() override = default;
+
+  bool Init(const CNodePtr &kernel_node) override {
+    // get bond_numbers
+    kernel_node_ = kernel_node;
+    atom_numbers = static_cast<int>(GetAttr<int64_t>(kernel_node, "atom_numbers"));
+    learning_rate = static_cast<float>(GetAttr<float>(kernel_node, "learning_rate"));
+    InitSizeLists();
+    return true;
+  }
+
+  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 {
+    auto crd = GetDeviceAddress<float>(inputs, 0);
+    auto frc = GetDeviceAddress<float>(inputs, 1);
+
+    MDIterationGradientDescent(atom_numbers, crd, frc, learning_rate, reinterpret_cast<cudaStream_t>(stream_ptr));
+    return true;
+  }
+
+ protected:
+  void InitSizeLists() override {
+    input_size_list_.push_back(atom_numbers * 3 * sizeof(T));
+    input_size_list_.push_back(atom_numbers * 3 * sizeof(T));
+
+    output_size_list_.push_back(sizeof(T));
+  }
+
+ private:
+  std::vector<size_t> input_size_list_;
+  std::vector<size_t> output_size_list_;
+  std::vector<size_t> workspace_size_list_;
+  int atom_numbers;
+  float learning_rate;
+};
+}  // namespace kernel
+}  // namespace mindspore
+#endif

mindspore/ccsrc/backend/kernel_compiler/gpu/sponge/nvtit/md_iteration_leap_frog_with_max_vel_kernel.cc

@@ -0,0 +1,31 @@
+/**
+ * Copyright 2021 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/sponge/nvtit/md_iteration_leap_frog_with_max_vel_kernel.h"
+
+namespace mindspore {
+namespace kernel {
+MS_REG_GPU_KERNEL_ONE(MDIterationLeapFrogWithMaxVel,
+                      KernelAttr()
+                        .AddInputAttr(kNumberTypeFloat32)
+                        .AddInputAttr(kNumberTypeFloat32)
+                        .AddInputAttr(kNumberTypeFloat32)
+                        .AddInputAttr(kNumberTypeFloat32)
+                        .AddInputAttr(kNumberTypeFloat32)
+                        .AddOutputAttr(kNumberTypeFloat32),
+                      MDIterationLeapFrogWithMaxVelGpuKernel, float)
+}  // namespace kernel
+}  // namespace mindspore

mindspore/ccsrc/backend/kernel_compiler/gpu/sponge/nvtit/md_iteration_leap_frog_with_max_vel_kernel.h

@@ -0,0 +1,86 @@
+/**
+ * Copyright 2021 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_MD_ITERATION_LEAP_FROG_WITH_MAX_VEL_KERNEL_H_
+#define MINDSPORE_CCSRC_BACKEND_KERNEL_COMPILER_GPU_MD_ITERATION_LEAP_FROG_WITH_MAX_VEL_KERNEL_H_
+
+#include "backend/kernel_compiler/gpu/cuda_impl/sponge/nvtit/md_iteration_leap_frog_with_max_vel_impl.cuh"
+#include <cuda_runtime_api.h>
+#include <map>
+#include <string>
+#include <vector>
+
+#include "backend/kernel_compiler/gpu/gpu_kernel.h"
+#include "backend/kernel_compiler/gpu/gpu_kernel_factory.h"
+#include "runtime/device/gpu/cuda_common.h"
+
+namespace mindspore {
+namespace kernel {
+template <typename T>
+class MDIterationLeapFrogWithMaxVelGpuKernel : public GpuKernel {
+ public:
+  MDIterationLeapFrogWithMaxVelGpuKernel() {}
+  ~MDIterationLeapFrogWithMaxVelGpuKernel() override = default;
+
+  bool Init(const CNodePtr &kernel_node) override {
+    // get bond_numbers
+    kernel_node_ = kernel_node;
+    atom_numbers = static_cast<int>(GetAttr<int64_t>(kernel_node, "atom_numbers"));
+    dt = static_cast<float>(GetAttr<float>(kernel_node, "dt"));
+    max_velocity = static_cast<float>(GetAttr<float>(kernel_node, "max_velocity"));
+    InitSizeLists();
+    return true;
+  }
+
+  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 {
+    auto vel = GetDeviceAddress<float>(inputs, 0);
+    auto crd = GetDeviceAddress<float>(inputs, 1);
+    auto frc = GetDeviceAddress<float>(inputs, 2);
+    auto acc = GetDeviceAddress<float>(inputs, 3);
+    auto inverse_mass = GetDeviceAddress<float>(inputs, 4);
+
+    MDIterationLeapFrogWithMaxVelocity(atom_numbers, vel, crd, frc, acc, inverse_mass, dt, max_velocity,
+                                       reinterpret_cast<cudaStream_t>(stream_ptr));
+    return true;
+  }
+
+ protected:
+  void InitSizeLists() override {
+    input_size_list_.push_back(atom_numbers * 3 * sizeof(T));
+    input_size_list_.push_back(atom_numbers * 3 * sizeof(T));
+    input_size_list_.push_back(atom_numbers * 3 * sizeof(T));
+    input_size_list_.push_back(atom_numbers * 3 * sizeof(T));
+    input_size_list_.push_back(atom_numbers * sizeof(T));
+
+    output_size_list_.push_back(sizeof(T));
+  }
+
+ private:
+  std::vector<size_t> input_size_list_;
+  std::vector<size_t> output_size_list_;
+  std::vector<size_t> workspace_size_list_;
+  int atom_numbers;
+  float dt;
+  float max_velocity;
+};
+}  // namespace kernel
+}  // namespace mindspore
+#endif