aclnnSiluBackward

📄 View Source Code

Applicable Products

Function

• Description: Backward propagation of aclnnSilu. Computes the gradient input of SiLU based on the backward propagation gradient of SiLU and the forward output.

• Formula:

  $$\sigma(x) = {\frac{1} {1+{e}^{-x}}}$$ $$s(x) = x\sigma(x)$$ $$s^\prime(x) = \sigma(x)(1+x-x\sigma(x))$$ $$gradInput = gradOutput * s^\prime(x)$$

where $\sigma(x)$ is the sigmoid function, $s(x)$ is the SiLU function, and $s^\prime(x)$ is the derivative of the SiLU function.

Function Prototype

Each operator is divided into a two-phase API. The "aclnnSiluBackwardGetWorkspaceSize" API must be called first to obtain the workspace size required for computation and the executor that encapsulates the operator computation flow. Then, the "aclnnSiluBackward" API is called to perform the computation.

aclnnStatus aclnnSiluBackwardGetWorkspaceSize(
  const aclTensor*    gradOutput,
  const aclTensor*    self,
  aclTensor*          gradInput,
  uint64_t*           workspaceSize,
  aclOpExecutor**     executor)

aclnnStatus aclnnSiluBackward(
  void*          workspace,
  uint64_t       workspaceSize,
  aclOpExecutor* executor,
  aclrtStream    stream)

aclnnSiluBackwardGetWorkspaceSize

• Parameters:

  Parameter	Input/Output	Description	Instruction	Data Type	Data Format	Dimension (Shape)	Non-contiguous Tensor
  gradOutput (aclTensor*)	Input	Represents the input gradient. Corresponds to gradOutput in the formula.	Supports empty tensors. gradOutput, self, and gradInput have the same data type and shape. The shapes of gradOutput, self, and gradInput satisfy the broadcast relationship.	BFLOAT16, FLOAT16, FLOAT	ND	1-8	√
  self (aclTensor*)	Input	Represents the input data. Corresponds to x in the formula and the input parameter of the forward pass.	Empty tensors are supported. The data types and shapes of gradOutput, self, and gradInput are consistent. The shapes of gradOutput, self, and gradInput satisfy the broadcast relationship.	BFLOAT16, FLOAT16, FLOAT	ND	1-8	√
  gradInput (aclTensor*)	Output	Represents the gradient with respect to the input data self. Corresponds to gradInput in the formula.	gradOutput, self, and gradInput have the same data type and shape. The shapes of gradOutput, self, and gradInput satisfy the broadcast relationship.	BFLOAT16, FLOAT16, FLOAT	ND	1-8	√
  workspaceSize (uint64_t*)	Output	Returns the workspace size to be allocated on the Device side.	-	-	-	-	-
  executor (aclOpExecutor**)	Output	Returns the operator executor, which contains the operator computation flow.	-	-	-	-	-

  • Atlas training products: The data type supports FLOAT16 and FLOAT.

• Return value

  aclnnStatus: return code. For details, see aclnn Return Code.

  The first-phase API performs input parameter validation and returns an error in the following scenarios:

  Return Code	Error Code	Description
  ACLNN_ERR_PARAM_NULLPTR	161001	The passed gradOutput, self, or gradInput is a null pointer.
  ACLNN_ERR_PARAM_INVALID	161002	The data type of gradOutput, self, or gradInput is not within the supported range.
  		The data types of gradOutput, self, and gradInput are different or do not meet the requirements.
  		The shapes of gradOutput, self, and gradInput are different or do not satisfy the broadcast relationship.

aclnnSiluBackward

• Parameters:

  Parameter	Input/Output	Description
  workspace	Input	Address of the workspace memory applied for on the Device side.
  workspaceSize	Input	Size of the workspace applied for on the Device side, obtained from the first-phase API aclnnSiluBackwardGetWorkspaceSize.
  executor	Input	Operator executor, which contains the operator computation flow.
  stream	Input	Specifies the stream for task execution.

• Return value

  aclnnStatus: return code. For details, see aclnn Return Code.

Constraints

• Deterministic computation:

  • aclnnSiluBackward defaults to a deterministic implementation.

Example

The following provides example code for reference only. For details about compilation and running, see Compile and Run Samples.

#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_silu_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 shapeSize = 1;
  for (auto i : shape) {
    shapeSize *= i;
  }
  return shapeSize;
}

int Init(int32_t deviceId, aclrtStream* stream) {
  // Boilerplate for resource initialization.
  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 = 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);
  // Call aclrtMalloc to allocate device memory.
  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);
  // Call aclrtMemcpy to copy data from the host to the device memory.
  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);

  // Calculate the strides of the contiguous tensor.
  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];
  }

  // Call aclCreateTensor to create an 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. (Boilerplate) Initialize the device/stream. See the ACL API manual.
  // Set deviceId based on the actual device.
  int32_t deviceId = 0;
  aclrtStream stream;
  auto ret = Init(deviceId, &stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);

  // 2. Construct the input and output. Customize the construction based on the API.
  std::vector<int64_t> gradOutputShape = {2, 3};
  std::vector<int64_t> selfShape = {2, 3};
  std::vector<int64_t> gradInputShape = {2, 3};
  void* gradOutDeviceAddr = nullptr;
  void* selfDeviceAddr = nullptr;
  void* gradInputDeviceAddr = nullptr;
  aclTensor* gradOut = nullptr;
  aclTensor* self = nullptr;
  aclTensor* gradInput = nullptr;
  std::vector<float> gradOutHostData = {1, 1, 1, 1, 1, 1};
  std::vector<float> selfHostData = {1, 2, 3, 4, 5, 6};
  std::vector<float> gradInputHostData = {0, 0, 0, 0, 0, 0};

  // Create the gradOut aclTensor.
  ret = CreateAclTensor(gradOutHostData, gradOutputShape, &gradOutDeviceAddr, aclDataType::ACL_FLOAT, &gradOut);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  // Create the self aclTensor.
  ret = CreateAclTensor(selfHostData, selfShape, &selfDeviceAddr, aclDataType::ACL_FLOAT, &self);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  // Create the gradInput aclTensor.
  ret = CreateAclTensor(gradInputHostData, gradInputShape, &gradInputDeviceAddr, aclDataType::ACL_FLOAT, &gradInput);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  // 3. Call the CANN operator library API. Replace with the specific API name.
  int64_t dim = 0;
  uint64_t workspaceSize = 0;
  aclOpExecutor* executor;
  // Call the first-phase API of aclnnSiluBackward.
  ret = aclnnSiluBackwardGetWorkspaceSize(gradOut, self, gradInput, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnSiluBackwardGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);
  // Apply for device memory based on the workspaceSize calculated by the first-phase API.
  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);
  }
  // Call the second-phase API of aclnnSiluBackward.
  ret = aclnnSiluBackward(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnSiluBackward failed. ERROR: %d\n", ret); return ret);

  // 4. (Boilerplate) Synchronize and wait for task execution to complete.
  ret = aclrtSynchronizeStream(stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSynchronizeStream failed. ERROR: %d\n", ret); return ret);

  // 5. Obtain the output value and copy the result from the device memory to the host memory. Modify this based on the specific API definition.
  auto size = GetShapeSize(gradInputShape);
  std::vector<float> outData(size, 0);
  ret = aclrtMemcpy(outData.data(), outData.size() * sizeof(outData[0]), gradInputDeviceAddr,
                    size * sizeof(outData[0]), 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("out result[%ld] is: %f\n", i, outData[i]);
  }

  // 6. Release aclTensor and aclScalar. Modify based on the API definition as needed.
  aclDestroyTensor(gradOut);
  aclDestroyTensor(self);
  aclDestroyTensor(gradInput);

  // 7. Release device resources. Modify based on the API definition as needed.
  aclrtFree(gradOutDeviceAddr);
  aclrtFree(selfDeviceAddr);
  aclrtFree(gradInputDeviceAddr);
  if (workspaceSize > 0) {
    aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();
  return 0;
}