aclnnMaxUnpool3dBackward

Applicable Products

Product	Supported
Ascend 950PR/Ascend 950DT	√
Atlas A3 training products/Atlas A3 inference products	√
Atlas A2 training products/Atlas A2 inference products	√
Atlas 200I/500 A2 inference products	×
Atlas inference products	×
Atlas training products	√

Function

Operator function: backward propagation of the inverse operation of MaxPool3d, aclnnMaxUnpool3d. Fills the elements of gradOutput into out according to the indices.
Formula:
- When the input is 4-dimensional, the dimensions represent (N, D, H, W):
$out[N][i] = gradOutput[N][indices[N][i]]$
- When the input is 5-dimensional, the dimensions represent (N, C, D, H, W):
$out[N][C][i] = gradOutput[N][C][indices[N][C][i]]$
where out, gradOutput, and indices are obtained by merging the last two axes into one axis and then reshaping, and i ∈ [0, DH W).

Function Prototype

Each operator is divided into a two-phase API. First call the "aclnnMaxUnpool3dBackwardGetWorkspaceSize" API to obtain the workspace size required for computation and the executor that contains the operator computation process, and then call the "aclnnMaxUnpool3dBackward" API to perform the computation.

aclnnStatus aclnnMaxUnpool3dBackwardGetWorkspaceSize(
  const aclTensor*     gradOutput,
  const aclTensor*     self,
  const aclTensor*     indices,
  const aclIntArray*   outputSize,
  const aclIntArray*   stride,
  const aclIntArray*   padding,
  aclTensor*           out,
  uint64_t*            workspaceSize,
  aclOpExecutor**      executor)

aclnnStatus aclnnMaxUnpool3dBackward(
  void*          workspace,
  uint64_t       workspaceSize,
  aclOpExecutor* executor,
  aclrtStream    stream)

aclnnMaxUnpool3dBackwardGetWorkspaceSize

Parameters

Parameter	Input/Output	Description	Instruction	Data Type	Data Format	Dimensions (shape)	Non-contiguous Tensor
gradOutput (aclTensor*)	Input	gradOutput in the formula.	The data type must be consistent with self and out. The dimensions represent (N, outputSize[0], outputSize[1], outputSize[2]) or (N, C, outputSize[0], outputSize[1], outputSize[2]). When the dimension is 4, the axes represent N, D, H, and W in sequence. When the dimension is 5, the axes represent N, C, D, H, and W in sequence.	FLOAT, FLOAT16, INT16, INT32, INT64, INT8, UINT8, DOUBLE	ND, NCDHW	4-5	√
self (aclTensor*)	Input	-	The data type must be consistent with gradOutput and out. The dimensions represent (N, D, H, W) or (N, C, D, H, W). The dimensions must be consistent with gradOutput, and the shape must be consistent with indices and out.	FLOAT, FLOAT16, INT16, INT32, INT64, INT8, UINT8, DOUBLE	ND, NCDHW	4-5	√
indices (aclTensor*)	Input	Indices in the formula, representing the index positions of the input gradOutput elements in the output result.	The shape must be consistent with self.	INT64	ND, NCDHW	4-5	√
outputSize (aclIntArray*)	Input	Indicates the spatial size of the output in the D, H, and W dimensions.	The size is 3.	INT64	-	-	-
stride (aclIntArray*)	Input	Specifies the step size of the max pooling window in the D, H, and W dimensions.	The size is 3.	INT64	-	-	-
padding (aclIntArray*)	Input	Specifies the padding values for the max pooling window in the D, H, and W dimensions.	The size is 3.	INT64	-	-	-
out (aclTensor*)	Output	The out in the formula.	The data type is the same as that of gradOutput and self. The shape must be the same as that of self and indices.	FLOAT, FLOAT16, INT16, INT32, INT64, INT8, UINT8, DOUBLE	ND, NCDHW	4-5	×
workspaceSize (uint64_t*)	Output	Returns the workspace size to be requested on the Device side.	-	-	-	-	-
executor (aclOpExecutor**)	Output	Returns the operator executor, which contains the operator computation flow.	-	-	-	-	-

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 Value	Error Code	Description
ACLNN_ERR_PARAM_NULLPTR	161001	The passed gradOutput, self, indices, outputSize, stride, padding, or out is a null pointer.
ACLNN_ERR_PARAM_INVALID	161002	The out is a non-contiguous tensor.
		The data type of gradOutput, self, indices, or out is not within the supported range.
		The data types of gradOutput, self, and out are inconsistent.
		The dimension of self is not 4D or 5D.
		The dimensions of self, indices, and out are inconsistent.
		The shapes of self, indices, and out are inconsistent.
		The size of self in each dimension except the N dimension is not greater than 0.
		The size of outputSize, stride, or padding is not equal to 3.
		An element value of outputSize or stride is not greater than 0.
		The product of the three elements of outputSize is less than the product of the sizes of self in the D, H, and W dimensions.
		The size of gradOutput in the D, H, and W dimensions is not equal to the three element values in outputSize.
		The dimensions of gradOutput and self are different.
		When self is 4-dimensional, the sizes of gradOutput and self in the N dimension are inconsistent.
		When self is 5-dimensional, the sizes of gradOutput and self in the C or N dimension are inconsistent.

aclnnMaxUnpool3dBackward

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 through the first-phase API aclnnLogAddExpGetWorkspaceSize.
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:
- aclnnMaxUnpool3dBackward 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_max_unpool3d_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 apply for device-side 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 a 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 the aclCreateTensor API 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.
  // Fill in the 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> selfShape = {1, 1, 4, 4};
  std::vector<int64_t> indicesShape = {1, 1, 4, 4};
  std::vector<int64_t> gradShape = {1, 1, 4, 4};
  std::vector<int64_t> outShape = {1, 1, 4, 4};
  void* gradDeviceAddr = nullptr;
  void* selfDeviceAddr = nullptr;
  void* indicesDeviceAddr = nullptr;
  void* outDeviceAddr = nullptr;
  aclTensor* grad = nullptr;
  aclTensor* self = nullptr;
  aclTensor* out = nullptr;
  aclTensor* indices = nullptr;
  std::vector<float> gradHostData = {1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1};
  std::vector<float> selfHostData = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15};
  std::vector<float> outHostData = {0, 0, 0, 0.0, 0, 0, 0, 0, 0, 0, 0, 0.0, 0, 0, 0, 0};
  std::vector<int64_t> indicesHostData = {0, 0, 0, 3, 0, 0, 0, 8, 0, 0, 0, 11, 0, 0, 0, 13};
  // Create the grad aclTensor.
  ret = CreateAclTensor(gradHostData, gradShape, &gradDeviceAddr, aclDataType::ACL_FLOAT, &grad);
  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 indices aclTensor.
  ret = CreateAclTensor(indicesHostData, indicesShape, &indicesDeviceAddr, aclDataType::ACL_INT64, &indices);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  // Create the out aclTensor.
  ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_FLOAT, &out);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  std::vector<int64_t> arraySize1 = {1, 4, 4};
  const aclIntArray *outputSize = aclCreateIntArray(arraySize1.data(), arraySize1.size());
  CHECK_RET(outputSize != nullptr, return ACL_ERROR_INTERNAL_ERROR);

  std::vector<int64_t> arraySize2 = {1, 2, 3};
  const aclIntArray *stride = aclCreateIntArray(arraySize2.data(), arraySize2.size());
  CHECK_RET(stride != nullptr, return ACL_ERROR_INTERNAL_ERROR);
  const aclIntArray *padding = aclCreateIntArray(arraySize2.data(), arraySize2.size());
  CHECK_RET(padding != nullptr, return ACL_ERROR_INTERNAL_ERROR);

  // 3. Call the CANN operator library API. Replace with the actual API name.
  uint64_t workspaceSize = 0;
  aclOpExecutor* executor;
  // Call the first-phase API of aclnnMaxUnpool3dBackward.
  ret = aclnnMaxUnpool3dBackwardGetWorkspaceSize(grad, self, indices, outputSize, stride, padding, out, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnMaxUnpool3dBackwardGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);
  // Allocate 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 aclnnMaxUnpool3dBackward.
  ret = aclnnMaxUnpool3dBackward(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnMaxUnpool3dBackward 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 based on the specific API definition.
  auto size = GetShapeSize(outShape);
  std::vector<float> outData(size, 0);
  ret = aclrtMemcpy(outData.data(), outData.size() * sizeof(outData[0]), outDeviceAddr, 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[%ld] is: %f\n", i, outData[i]);
  }

  // 6. Release aclTensor and aclScalar. Modify based on the specific API definition.
  aclDestroyTensor(grad);
  aclDestroyTensor(self);
  aclDestroyTensor(out);
  aclDestroyTensor(indices);
  aclDestroyIntArray(outputSize);
  aclDestroyIntArray(stride);
  aclDestroyIntArray(padding);

  // 7. Release device resources.
  aclrtFree(gradDeviceAddr);
  aclrtFree(selfDeviceAddr);
  aclrtFree(indicesDeviceAddr);
  aclrtFree(outDeviceAddr);
  if (workspaceSize > 0) {
    aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();
  return 0;
}