aclnnBatchNormBackward

#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_batch_norm_backward.h"

#define CHECK_RET(cond, return_expr) \
  do {                               \
    if (!(cond)) {                   \
      return_expr;                   \
    }                                \
  } while (0)

#define LOG_PRINT(message, ...)     \
  do {                              \
    printf(message, ##__VA_ARGS__); \
  } while (0)

int64_t GetShapeSize(const std::vector<int64_t>& shape) {
  int64_t shape_size = 1;
  for (auto i : shape) {
    shape_size *= i;
  }
  return shape_size;
}

int Init(int32_t deviceId, aclrtContext* context, aclrtStream* stream) {
  // 固定写法，AscendCL初始化
  auto ret = aclInit(nullptr);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclInit failed. ERROR: %d\n", ret); return ret);
  ret = aclrtSetDevice(deviceId);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSetDevice failed. ERROR: %d\n", ret); return ret);
  ret = aclrtCreateContext(context, deviceId);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtCreateContext failed. ERROR: %d\n", ret); return ret);
  ret = aclrtSetCurrentContext(*context);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSetCurrentContext failed. ERROR: %d\n", ret); return ret);
  ret = aclrtCreateStream(stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtCreateStream failed. ERROR: %d\n", ret); return ret);
  return 0;
}

template <typename T>
int CreateAclTensor(const std::vector<T>& hostData, const std::vector<int64_t>& shape, void** deviceAddr,
                    aclDataType dataType, aclTensor** tensor) {
  auto size = GetShapeSize(shape) * sizeof(T);
  // 调用aclrtMalloc申请device侧内存
  auto ret = aclrtMalloc(deviceAddr, size, ACL_MEM_MALLOC_HUGE_FIRST);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtMalloc failed. ERROR: %d\n", ret); return ret);

  // 调用aclrtMemcpy将Host侧数据拷贝到device侧内存上
  ret = aclrtMemcpy(*deviceAddr, size, hostData.data(), size, ACL_MEMCPY_HOST_TO_DEVICE);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtMemcpy failed. ERROR: %d\n", ret); return ret);

  // 计算连续tensor的strides
  std::vector<int64_t> strides(shape.size(), 1);
  for (int64_t i = shape.size() - 2; i >= 0; i--) {
    strides[i] = shape[i + 1] * strides[i + 1];
  }

  // 调用aclCreateTensor接口创建aclTensor
  *tensor = aclCreateTensor(shape.data(), shape.size(), dataType, strides.data(), 0, aclFormat::ACL_FORMAT_ND,
                            shape.data(), shape.size(), *deviceAddr);
  return 0;
}

int main() {
  // 1. （固定写法）device/context/stream初始化, 参考AscendCL对外接口列表
  // 根据自己的实际device填写deviceId
  int32_t deviceId = 0;
  aclrtContext context;
  aclrtStream stream;
  auto ret = Init(deviceId, &context, &stream);
  // check根据自己的需要处理
  CHECK_RET(ret == 0, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);

  // 2. 构造输入与输出，需要根据API的接口自定义构造
  std::vector<int64_t> inputShape = {2, 3, 2};
  std::vector<int64_t> meanShape = {3};
  void* gradOutDeviceAddr = nullptr;
  void* inputDeviceAddr = nullptr;
  void* weightDeviceAddr = nullptr;
  void* runningMeanDeviceAddr = nullptr;
  void* runningVarDeviceAddr = nullptr;
  void* saveMeanDeviceAddr = nullptr;
  void* saveInvstdDeviceAddr = nullptr;
  void* gradInputDeviceAddr = nullptr;
  void* gradWeightDeviceAddr = nullptr;
  void* gradBiasDeviceAddr = nullptr;
  aclTensor* gradOut = nullptr;
  aclTensor* input = nullptr;
  aclTensor* weight = nullptr;
  aclTensor* runningMean = nullptr;
  aclTensor* runningVar = nullptr;
  aclTensor* saveMean = nullptr;
  aclTensor* saveInvstd = nullptr;
  aclTensor* gradInput = nullptr;
  aclTensor* gradWeight = nullptr;
  aclTensor* gradBias = nullptr;
  std::vector<float> inputHostData = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11};
  std::vector<float> zeroHostData = {0, 0, 0};
  std::vector<float> oneHostData = {1, 1, 1};
  // 创建gradOut aclTensor
  ret = CreateAclTensor(inputHostData, inputShape, &gradOutDeviceAddr, aclDataType::ACL_FLOAT, &gradOut);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建input aclTensor
  ret = CreateAclTensor(inputHostData, inputShape, &inputDeviceAddr, aclDataType::ACL_FLOAT, &input);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建weight aclTensor
  ret = CreateAclTensor(oneHostData, meanShape, &weightDeviceAddr, aclDataType::ACL_FLOAT, &weight);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建runningMean aclTensor
  ret = CreateAclTensor(zeroHostData, meanShape, &runningMeanDeviceAddr, aclDataType::ACL_FLOAT, &runningMean);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建runningVar aclTensor
  ret = CreateAclTensor(oneHostData, meanShape, &runningVarDeviceAddr, aclDataType::ACL_FLOAT, &runningVar);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建saveMean aclTensor
  ret = CreateAclTensor(zeroHostData, meanShape, &saveMeanDeviceAddr, aclDataType::ACL_FLOAT, &saveMean);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建saveInvstd aclTensor
  ret = CreateAclTensor(zeroHostData, meanShape, &saveInvstdDeviceAddr, aclDataType::ACL_FLOAT, &saveInvstd);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建gradInput aclTensor
  ret = CreateAclTensor(inputHostData, inputShape, &gradInputDeviceAddr, aclDataType::ACL_FLOAT, &gradInput);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建gradWeight aclTensor
  ret = CreateAclTensor(zeroHostData, meanShape, &gradWeightDeviceAddr, aclDataType::ACL_FLOAT, &gradWeight);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // 创建gradBias aclTensor
  ret = CreateAclTensor(zeroHostData, meanShape, &gradBiasDeviceAddr, aclDataType::ACL_FLOAT, &gradBias);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  std::array<bool, 3> value = {true, true, true};
  auto outputMask = aclCreateBoolArray(value.data(), value.size());

  // 3.调用CANN算子库API，需要修改为具体的算子接口
  uint64_t workspaceSize = 0;
  aclOpExecutor* executor;
  // 调用aclnnBatchNormBackward第一段接口
  ret = aclnnBatchNormBackwardGetWorkspaceSize(gradOut, input, weight, runningMean, runningVar, saveMean, saveInvstd,
                                               true, 1e-5, outputMask, gradInput, gradWeight, gradBias,
                                               &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnBatchNormBackwardGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);
  // 根据第一段接口计算出的workspaceSize申请device内存
  void* workspaceAddr = nullptr;
  if (workspaceSize > 0) {
    ret = aclrtMalloc(&workspaceAddr, workspaceSize, ACL_MEM_MALLOC_HUGE_FIRST);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("allocate workspace failed. ERROR: %d\n", ret); return ret;);
  }
  // 调用aclnnBatchNormBackward第二段接口
  ret = aclnnBatchNormBackward(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnBatchNormBackward failed. ERROR: %d\n", ret); return ret);

  // 4. （固定写法）同步等待任务执行结束
  ret = aclrtSynchronizeStream(stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSynchronizeStream failed. ERROR: %d\n", ret); return ret);

  // 5. 获取输出的值，将device侧内存上的结果拷贝至Host侧，需要根据具体API的接口定义修改
  auto size = GetShapeSize(inputShape);
  std::vector<float> resultData(size, 0);
  ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), gradInput, size * sizeof(float),
                    ACL_MEMCPY_DEVICE_TO_HOST);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("copy result from device to host failed. ERROR: %d\n", ret); return ret);
  for (int64_t i = 0; i < size; i++) {
    LOG_PRINT("result[%ld] is: %f\n", i, resultData[i]);
  }

  // 6. 释放aclTensor和aclScalar，需要根据具体API的接口定义修改
  aclDestroyTensor(gradOut);
  aclDestroyTensor(input);
  aclDestroyTensor(weight);
  aclDestroyTensor(runningMean);
  aclDestroyTensor(runningVar);
  aclDestroyTensor(saveMean);
  aclDestroyTensor(saveInvstd);
  aclDestroyTensor(gradInput);
  aclDestroyTensor(gradWeight);
  aclDestroyTensor(gradBias);
  return 0;
}