aclnnLeakyRelu&aclnnInplaceLeakyRelu

Supported Products

Product	Supported
Atlas A3 training series products/Atlas A3 inference series products	√
Atlas A2 training products/Atlas A2 inference products	√
Atlas 200I/500 A2 inference products	×
Atlas inference series products	×
Atlas training series products	√

Function

Description: Activation function, which is used to solve the problem that the output of the Relu function is 0 when the input is less than 0 to prevent neurons from failing to update parameters.
Formula: $out = max(0,self) + negativeSlope * min(0,self)$

Prototype

aclnnLeakyRelu and aclnnInplaceLeakyRelu implement the same function in different ways. Select a proper operator based on your requirements.
- aclnnLeakyRelu: An output tensor object needs to be created to store the computation result.
- aclnnInplaceLeakyRelu: No output tensor object needs to be created, and the computation result is stored in the memory of the input tensor.
Each operator has two-phase API calls. First, aclnnLeakyReluGetWorkspaceSize or aclnnInplaceLeakyReluGetWorkspaceSize is called to obtain the workspace size required for computation and the executor that contains the operator computation process. Then, aclnnLeakyRelu or aclnnInplaceLeakyRelu is called to perform computation.

aclnnStatus aclnnLeakyReluGetWorkspaceSize(
  const aclTensor *self,
  const aclScalar *negativeSlope,
  aclTensor       *out,
  uint64_t        *workspaceSize,
  aclOpExecutor  **executor)

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

aclnnStatus aclnnInplaceLeakyReluGetWorkspaceSize(
  aclTensor       *selfRef,
  const aclScalar *negativeSlope,
  uint64_t        *workspaceSize,
  aclOpExecutor  **executor)

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

aclnnLeakyReluGetWorkspaceSize

Parameters:

Name	Input/Output	Description	Usage Notes	Data Type	Data Format	Dimension (Shape)	Non-contiguous Tensor
self	Input	Input of the LeakyRelu activation function, corresponding to self in the formula.	The shape can be 0 to 8 dimensions. The shape must be the same as that of out. Empty tensors are supported.	FLOAT, FLOAT16, BFLOAT16, DOUBLE	ND	0-8	√
negativeSlope	Input	Slope when self is less than 0, corresponding to negativeSlope in the formula.	-	FLOAT	-	-	-
out	Output	Output of the LeakyRelu activation function.	The data type of out must be convertible from that of self (see Conversion Relationship).	FLOAT, FLOAT16, BFLOAT16, DOUBLE	ND	0-8	√
workspaceSize	Output	Size of the workspace to be allocated on the device.	-	-	-	-	-
executor	Output	Operator executor, containing the operator computation process.	-	-	-	-	-

Atlas training series products: FLOAT, FLOAT16, and DOUBLE.

Returns:

aclnnStatus: status code. For details, see aclnn Return Codes.

The first-phase API implements input parameter verification. The following errors may be thrown:

Return	Error Code	Description
ACLNN_ERR_PARAM_NULLPTR	161001	The passed self, negativeSlope, or out is a null pointer.
ACLNN_ERR_PARAM_INVALID	161002	The data type of self is not supported.
		The shape of self has more than eight dimensions.
		The shape of out is inconsistent with that of self.

aclnnLeakyRelu

Parameters:

Name	Input/Output	Description
workspace	Input	Address of the workspace to be allocated on the device.
workspaceSize	Input	Size of the workspace to be allocated on the device, which is obtained by the first-phase API aclnnLeakyReluGetWorkspaceSize.
executor	Input	Operator executor, containing the operator computation process.
stream	Input	Stream for executing the task.

Returns:

aclnnStatus: status code. For details, see aclnn Return Codes.

aclnnInplaceLeakyReluGetWorkspaceSize

Parameters:

Name	Input/Output	Description	Usage Notes	Data Type	Data Format	Dimension (Shape)	Non-contiguous Tensor
selfRef	Input and output	self and out in the formula.	Empty tensors are supported.	FLOAT, FLOAT16, BFLOAT16, DOUBLE	ND	0-8	√
negativeSlope	Input	Slope when self is less than 0, corresponding to negativeSlope in the formula.	-	FLOAT	-	-	-
workspaceSize	Output	Size of the workspace to be allocated on the device.	-	-	-	-	-
executor	Output	Operator executor, containing the operator computation process.	-	-	-	-	-

Atlas training series products: The data type can be FLOAT, FLOAT16, or DOUBLE.

Returns:

aclnnStatus: status code. For details, see aclnn Return Codes.

The first-phase API implements input parameter verification. The following errors may be thrown:

Return	Error Code	Description
ACLNN_ERR_PARAM_NULLPTR	161001	The passed selfRef or negativeSlope is a null pointer.
ACLNN_ERR_PARAM_INVALID	161002	The data type of selfRef is not supported.

aclnnInplaceLeakyRelu

Parameters:

Name	Input/Output	Description
workspace	Input	Address of the workspace to be allocated on the device.
workspaceSize	Input	Size of the workspace to be allocated on the device, which is obtained by the first-phase API aclnnInplaceLeakyReluGetWorkspaceSize.
executor	Input	Operator executor, containing the operator computation process.
stream	Input	Stream for executing the task.

Returns:

aclnnStatus: status code. For details, see aclnn Return Codes.

Constraints

Deterministic compute:
- aclnnLeakyRelu&aclnnInplaceLeakyRelu defaults to a deterministic implementation.
If negativeSlope is an integer and self is of the FLOAT type, precision loss may occur when negativeSlope is greater than 2^24 or less than -2^24. Similarly, if self is of the FLOAT16 type, precision loss may occur when negativeSlope is greater than 2^11 or less than -2^11.

Example

The following example is for reference only. For details, see Compilation and Running Sample.

aclnnLeakyRelu sample code:

#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_leaky_relu.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) {
  // (Fixed writing) Initialize resources.
  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 memory on the device.
  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 the data on the host to the memory on the device.
  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);

  // Compute 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. (Fixed writing) Initialize the device and stream. For details, see the ACL API manual.
  // Set the device ID in use.
  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 based on the API.
  std::vector<int64_t> selfShape = {4};
  std::vector<int64_t> outShape = {4};
  void* selfDeviceAddr = nullptr;
  void* outDeviceAddr = nullptr;
  aclTensor* self = nullptr;
  aclScalar* negativeSlope = nullptr;
  aclTensor* out = nullptr;
  std::vector<float> selfHostData = {1, 2, 3, 4};
  std::vector<float> outHostData = {0, 0, 0, 0};
  float negativeSlopeValue = 0.01f;
  // Create a self aclTensor.
  ret = CreateAclTensor(selfHostData, selfShape, &selfDeviceAddr, aclDataType::ACL_FLOAT, &self);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create a negativeSlope aclScalar.
  negativeSlope = aclCreateScalar(&negativeSlopeValue, aclDataType::ACL_FLOAT);
  CHECK_RET(negativeSlope != nullptr, LOG_PRINT("negativeSlope is null!\n", negativeSlope); return false);
  // Create an out aclTensor.
  ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_FLOAT, &out);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  // 3. Call the CANN operator library API. Modify the API as required.
  uint64_t workspaceSize = 0;
  aclOpExecutor* executor;
  // Call the first-phase API of aclnnLeakyRelu.
  ret = aclnnLeakyReluGetWorkspaceSize(self, negativeSlope, out, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnLeakyReluGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);
  // Allocate device memory based on workspaceSize computed 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 aclnnLeakyRelu.
  ret = aclnnLeakyRelu(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnLeakyRelu failed. ERROR: %d\n", ret); return ret);

  // 4. (Fixed writing) Wait until the task execution is 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. Modify the configuration based on the API definition.
  auto size = GetShapeSize(outShape);
  std::vector<float> resultData(size, 0);
  ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), outDeviceAddr,
                    size * sizeof(resultData[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("result[%ld] is: %f\n", i, resultData[i]);
  }

  // 6. Release aclTensor and aclScalar. Modify the configuration based on the API definition.
  aclDestroyTensor(self);
  aclDestroyScalar(negativeSlope);
  aclDestroyTensor(out);
    
  // 7. Release device resources. Modify the configuration based on the API definition.
  aclrtFree(selfDeviceAddr);
  aclrtFree(outDeviceAddr);
  if (workspaceSize > 0) {
    aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();
  return 0;
}

aclnnInplaceLeakyRelu sample code:

#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_leaky_relu.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) {
  // (Fixed writing) Initialize resources.
  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 memory on the device.
  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 the data on the host to the memory on the device.
  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);

  // Compute 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. (Fixed writing) Initialize the device and stream. For details, see the ACL API manual.
  // Set the device ID in use.
  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 based on the API.
  std::vector<int64_t> selfRefShape = {4};
  void* selfRefDeviceAddr = nullptr;
  aclTensor* selfRef = nullptr;
  aclScalar* negativeSlope = nullptr;
  std::vector<float> selfRefHostData = {1, 2, 3, 4};
  float negativeSlopeValue = 0.01f;
  // Create a selfRef aclTensor.
  ret = CreateAclTensor(selfRefHostData, selfRefShape, &selfRefDeviceAddr, aclDataType::ACL_FLOAT, &selfRef);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create a negativeSlope aclScalar.
  negativeSlope = aclCreateScalar(&negativeSlopeValue, aclDataType::ACL_FLOAT);
  CHECK_RET(negativeSlope != nullptr, LOG_PRINT("negativeSlope is null!\n", negativeSlope); return false);

  // 3. Call the CANN operator library API. Modify the API as required.
  uint64_t workspaceSize = 0;
  aclOpExecutor* executor;
  // Call the first-phase API of aclnnInplaceLeakyRelu.
  ret = aclnnInplaceLeakyReluGetWorkspaceSize(selfRef, negativeSlope, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnInplaceLeakyReluGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);
  // Allocate device memory based on workspaceSize computed 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 aclnnInplaceLeakyRelu.
  ret = aclnnInplaceLeakyRelu(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnInplaceLeakyRelu failed. ERROR: %d\n", ret); return ret);

  // 4. (Fixed writing) Wait until the task execution is 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. Modify the configuration based on the API definition.
  auto size = GetShapeSize(selfRefShape);
  std::vector<float> resultData(size, 0);
  ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), selfRefDeviceAddr,
                    size * sizeof(resultData[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("result[%ld] is: %f\n", i, resultData[i]);
  }

  // 6. Release aclTensor and aclScalar. Modify the configuration based on the API definition.
  aclDestroyTensor(selfRef);
  aclDestroyScalar(negativeSlope);

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