aclnnAdaptiveMaxPool3dBackward

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 forward Adaptive Max Pooling operation. It backfills the gradient to the coordinates of the maximum value within each adaptive window, accumulating at identical coordinates.

Forward formula:

N (Batch) represents the batch size, C (Channels) represents the feature map channels, D (Depth) represents the feature map depth, H (Height) represents the feature map height, and W (Width) represents the feature map width. For the scenario where the input self has dimensions $[N,C,D,H,W]$ and the outputSize value is $[D_o,H_o,W_o]$ , the output dimensions are $[N,C,D_o,H_o,W_o]$ , the indices dimensions are $[N,C,D_o,H_o,W_o]$ , and the calculation formula for each element $(l,m,n)$ in the corresponding tensor is as follows:
$D^{l}_{left} = \lfloor(l*D)/D_o\rfloor$ $D^{l}_{right} = \lceil((l+1)*D)/D_o\rceil$ $H^{m}_{left} = \lfloor(m*H)/H_o\rfloor$ $H^{m}_{right} = \lceil((m+1)*H)/H_o\rceil$ $W^{n}_{left} = \lfloor(n*W)/W_o\rfloor$ $W^{n}_{right} = \lceil((n+1)*W)/W_o\rceil$ $output(N,C,l,m,n)=\underset {i \in [D^{l}_{left}, D^{l}_{right}],j\in [H^m_{left},H^m_{right}], k \in [W^n_{left},W^n_{right}] }{max} input(N,C,i,j,k)$ $indices(N,C,l,m,n)=\underset {i \in [D^{l}_{left}, D^{l}_{right}],j\in [H^m_{left},H^m_{right}], k \in [W^n_{left},W^n_{right}] }{argmax} input(N,C,i,j,k)$

Function Prototype

Each operator is divided into a two-phase API. First call the "aclnnAdaptiveMaxPool3dBackwardGetWorkspaceSize" API to obtain the workspace size required for computation and the executor that contains the operator computation process, and then call the "aclnnAdaptiveMaxPool3dBackward" API to perform the computation.

aclnnStatus aclnnAdaptiveMaxPool3dBackwardGetWorkspaceSize(
  const aclTensor   *gradOutput,
  const aclTensor   *self,
  const aclTensor   *indices,
  aclTensor         *gradInput,
  uint64_t          *workspaceSize,
  aclOpExecutor     **executor)

aclnnStatus aclnnAdaptiveMaxPool3dBackward(
  void          *workspace,
  uint64_t       workspaceSize,
  aclOpExecutor *executor,
  aclrtStream    stream)

aclnnAdaptiveMaxPool3dBackwardGetWorkspaceSize

Parameters:

Parameter	Input/Output	Description	Instruction	Data Type	Data Format	Dimension (Shape)	Non-contiguous Tensor
gradOutput	Input	Gradient of the current node.	The shape is consistent with the forward output.	FLOAT32, FLOAT16, BFLOAT16	ND	4-5	√
self	Input	The input tensor of the forward pass.	The data type must be the same as gradOutput	FLOAT32、FLOAT16、BFLOAT16	ND	4-5	√
indices	Input	Index positions of the maximum elements in the forward input.	The shape is the same as gradOutput	INT32	ND	4-5	√
gradInput	Output	Backward output tensor	The shape is consistent with self.	FLOAT32, FLOAT16, BFLOAT16	ND	4-5	√
workspaceSize	Output	Returns the workspace size to be allocated on the Device side.	-	-	-	-	-
executor	Output	Returns the op executor, which contains the operator computation process.	-	-	-	-	-

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 indices is a null pointer.
ACLNN_ERR_PARAM_INVALID	161002	The data types of gradOutput, self, indices, and gradInput are not within the supported range.
		The data formats of gradOutput, self, indices, and gradInput are not within the supported range.
		The shape of the input or output is not 4-dimensional or 5-dimensional.
		The shapes of gradOutput and indices are inconsistent, or the shapes of self and gradInput are inconsistent.
		depth * height * width exceeds the maximum int32 value, which is beyond the representable range of indices.

aclnnAdaptiveMaxPool3dBackward

Parameters:

Parameter	Input/Output	Description
workspace	Input	Specifies the workspace memory address applied on the Device side.
workspaceSize	Input	Specifies the workspace size applied on the Device side, which is obtained through the first-phase API aclnnAdaptiveMaxPool3dBackwardGetWorkspaceSize.
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:
- aclnnAdaptiveMaxPool3dBackward defaults to a non-deterministic implementation. Deterministic computation can be enabled through aclrtCtxSetSysParamOpt.
In non-divisible scenarios, the shape does not exceed 2 to the power of 24. In divisible scenarios, this restriction does not apply.

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_adaptive_max_pool3d_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 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 host data to 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 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 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);
  // Handle the check based on actual requirements.
  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> gradOutShape = {1, 1, 1, 1, 1};
  std::vector<int64_t> selfShape = {1, 1, 2, 2, 2};
  std::vector<int64_t> indicesShape = {1, 1, 1, 1, 1};
  std::vector<int64_t> gradInShape = {1, 1, 2, 2, 2};
  void* gradOutDeviceAddr = nullptr;
  void* selfDeviceAddr = nullptr;
  void* indicesDeviceAddr = nullptr;
  void* gradInDeviceAddr = nullptr;
  aclTensor* gradOut = nullptr;
  aclTensor* self = nullptr;
  aclTensor* indices = nullptr;
  aclTensor* gradIn = nullptr;
  std::vector<float> gradOutHostData = {0.4757};
  std::vector<float> selfHostData = {0.0850, -0.5147, -0.0212, -0.5654, -0.3222, 0.5847, 1.7510, 0.9954};
  std::vector<int32_t> indicesHostData = {6};
  std::vector<float> gradInHostData = {0, 0, 0, 0, 0, 0, 0, 0};

  // Create the gradOut aclTensor.
  ret = CreateAclTensor(gradOutHostData, gradOutShape, &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 indices aclTensor.
  ret = CreateAclTensor(indicesHostData, indicesShape, &indicesDeviceAddr, aclDataType::ACL_INT32, &indices);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create the gradIn aclTensor.
  ret = CreateAclTensor(gradInHostData, gradInShape, &gradInDeviceAddr, aclDataType::ACL_FLOAT, &gradIn);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  uint64_t workspaceSize = 0;
  aclOpExecutor* executor;

  // aclnnAdaptiveMaxPool3dBackward API call example
  // 3. Call the CANN operator library API. Replace with the actual API name.
  // Call the first-phase API aclnnAdaptiveMaxPool3dBackwardGetWorkspaceSize.
  ret = aclnnAdaptiveMaxPool3dBackwardGetWorkspaceSize(gradOut, self, indices, gradIn, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnAdaptiveMaxPool3dBackwardGetWorkspaceSize 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 aclnnAdaptiveMaxPool3dBackward.
  ret = aclnnAdaptiveMaxPool3dBackward(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnAdaptiveMaxPool3dBackward 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 side memory to the Host side. Modify this based on the specific API definition.
  auto size = GetShapeSize(gradInShape);
  std::vector<float> resultData(size, 0);
  ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), gradInDeviceAddr,
                    size * sizeof(resultData[0]), ACL_MEMCPY_DEVICE_TO_HOST);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("copy gradIn 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 this based on the specific API definition.
  aclDestroyTensor(gradOut);
  aclDestroyTensor(self);
  aclDestroyTensor(indices);
  aclDestroyTensor(gradIn);

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