aclnnUpsampleNearestExact3dBackward

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

Description: backward computation of aclnnUpsampleNearestExact3d.
Formula: $scalesD = outputSize[0] / inputSize[2]$ $scalesH = outputSize[1] / inputSize[3]$ $scalesW = outputSize[2] / inputSize[4]$ $gradInput(N, C, floor ( scalesD * ( D + 0.5 )), floor ( scalesH * ( H + 0.5 )), floor ( scalesW * ( W+ 0.5 ))) += gradOutput( N, C, D, H ,W)$

Function Prototype

Each operator is divided into a two-phase API. You must first call the "aclnnUpsampleNearestExact3dBackwardGetWorkspaceSize" API to obtain the input parameters and calculate the required workspace size based on the computation process, and then call the "aclnnUpsampleNearestExact3dBackward" API to perform the computation.

aclnnStatus aclnnUpsampleNearestExact3dBackwardGetWorkspaceSize(
  const aclTensor   *gradOut,
  const aclIntArray *outputSize,
  const aclIntArray *inputSize,
  double             scalesD,
  double             scalesH,
  double             scalesW,
  aclTensor         *gradInput,
  uint64_t          *workspaceSize,
  aclOpExecutor    **executor)

aclnnStatus aclnnUpsampleNearestExact3dBackward(
  void          *workspace,
  uint64_t       workspaceSize,
  aclOpExecutor *executor,
  aclrtStream    stream)

aclnnUpsampleNearestExact3dBackwardGetWorkspaceSize

Parameters

Parameter Name	Input/Output	Description	Usage Description	Data Type	Data Format	Dimension (shape)	Non-contiguous Tensor
gradOut (aclTensor*)	Input	Represents the gradient tensor for backward computation, corresponding to `gradOutput` in the formula.	Empty tensors are not supported. The value of each dimension of `gradOut` is less than or equal to (2^31-1).	FLOAT32, FLOAT16, BFLOAT16	NCDHW, NDHWC	5	√
outputSize (aclIntArray*)	Input	Represents the spatial size of the input `gradOut` in the D, H, and W dimensions, which equals `outputSize` in the corresponding formula.	size is 3, and each element is greater than zero.	INT64	-	-	-
inputSize (aclIntArray*)	Input	Represents the spatial size of the output `gradInput` in the N, C, D, H, and W dimensions, corresponding to `inputSize` in the formula.	The size is 5, and each element is greater than zero.	INT64	-	-	-
scalesD (double)	Input	Represents the depth dimension multiplier of the output `gradInput`, corresponding to `scalesD` in the corresponding formula.	The value is less than or equals 50.	-	-	-	-
scalesH (double)	Input	Represents the height dimension multiplier of the output `gradInput`, corresponding to `scalesH` in the formula.	The value is less than or equal to 50.	-	-	-	-
scalesW (double)	Input	Represents the width dimension multiplier of the output `gradInput`, corresponding to `scalesW` in the corresponding formula.	The value is less than or equal to 50.	-	-	-	-
gradInput (aclTensor*)	Output	Represents the output tensor of the backward computation, corresponding to the output `gradInput` in the formula.	Empty tensors are not supported. The data type, data format, and shape must be consistent with the input parameter `gradOut`. All dimensional values of gradInput must be less than or equal to (2^31-1).	FLOAT32, FLOAT16, BFLOAT16	NCDHW, NDHWC	5	√
workspaceSize (uint64_t*)	Output	Returns the workspace size to be applied on the Device side.	-	-	-	-	-
executor (aclOpExecutor**)	Output	Returns the op executor, which contains the operator computation process.	-	-	-	-	-

Return Value

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

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

Return Code	Error Code	Description
ACLNN_ERR_PARAM_NULLPTR	161001	The passed gradOut, outputSize, inputSize, or gradInput is a null pointer.
ACLNN_ERR_PARAM_INVALID	161002	The data type of gradOut is not within the supported range.
		The data types of gradOut and gradInput are inconsistent.
		The dimension of gradOut is not 5.
		The size of outputSize is not equal to 3.
		An element value in outputSize is not greater than 0.
		The size of inputSize is not equal to 5.
		The sizes of gradOut and inputSize are different in the N and C dimensions.
		The sizes of gradOut in the D, H, and W dimensions are inconsistent with outputSize[0], outputSize[1], and outputSize[2].
		The sizes of gradInput in the N and C dimensions are inconsistent with inputSize[0] and inputSize[1].
		The sizes of gradInput in the D, H, and W dimensions are inconsistent with inputSize[2], inputSize[3], and inputSize[4].
		The values of scalesD, scalesH, and scalesW do not meet the constraint requirements.

aclnnUpsampleNearestExact3dBackward

Parameters

Parameter Name	Input/Output	Description
workspace	Input	The workspace memory address applied on the Device side.
workspaceSize	Input	Workspace size applied on the Device side, obtained by the first-phase API aclnnUpsampleNearestExact3dBackwardGetWorkspaceSize.
executor	Input	Operator executor, which contains the operator computation flow.
stream	Input	Specifies the Stream for executing the task.

Return Value

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

Constraints

Shape constraints for parameters gradOut and gradInput:
- The value of each dimension is less than or equal to 2^20.
- The N axis and C axis of parameter gradInput remain consistent with those of gradOut.
- The memory usage must be less than 60 GB. The memory usage calculation formula is as follows: $N * C * (gradOut\_D * gradOut\_H * gradOut\_W + gradInput\_D * gradInput\_H * gradInput\_W + gradOut\_D * gradOut\_H * gradInput\_W + gradOut\_D * gradInput\_H * gradInput\_W) * sizeof(float) < 60 * 1024 * 1024 * 1024$ Where:
  - N represents the N axis of the input and output.
  - C represents the C axis of the input and output.
- N * C * gradOut_D * gradOut_H < 2^31
- gradInput_W * gradInput_H < 2^31
When the data format of the gradOut and gradInput parameters is not NCDHW or NDHWC, other input data formats are processed as NCDHW by default.
In the input data scaling scenario, the magnification factor must be less than or equal to 50, that is:
$outputSize\_D / depth D of the output shape <= 50$ $outputSize\_H / height H of the output shape <= 50$ $outputSize\_W / output shape width W <= 50$
The parameters inputSize, outputSize, scalesD, scalesH, and scalesW must satisfy the following constraints:
$outputSize\_D = floor(inputSize\_D * scalesD)$ $outputSize\_H = floor(inputSize\_H * scalesH)$ $outputSize\_W = floor(inputSize\_W * scalesW)$
Deterministic computation:
- aclnnUpsampleNearestExact3dBackward uses default deterministic implementation.

Call Example

The sample code is as follows, for reference only. For specific compilation and execution procedures, see Compile and Run Samples.

#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_upsample_nearest_exact3d_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, aclrtStream* stream)
{
    // Boilerplate: 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 memory on the device side.
    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 side 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 the aclCreateTensor API to create an aclTensor.
    *tensor = aclCreateTensor(
        shape.data(), shape.size(), dataType, strides.data(), 0, aclFormat::ACL_FORMAT_NCDHW, 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 your actual device.
    int32_t deviceId = 0;
    aclrtStream stream;
    auto ret = Init(deviceId, &stream);
    // Handle the check based on your own needs.
    CHECK_RET(ret == 0, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);
    // 2. Construct the input and output. Customize them based on the API.
    std::vector<int64_t> gradOutShape = {2, 2, 2, 2, 2};
    std::vector<int64_t> gradInputShape = {2, 2, 1, 1, 1};
    void* gradOutDeviceAddr = nullptr;
    void* gradInputDeviceAddr = nullptr;
    aclTensor* gradOut = nullptr;
    aclTensor* gradInput = nullptr;
    std::vector<float> gradOutHostData = {1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 16,
                                          17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32};
    std::vector<float> gradInputHostData = {2.0, 2, 2, 2};
    std::vector<int64_t> outputSizeData = {2, 2, 2};
    std::vector<int64_t> inputSizeData = {2, 2, 1, 1, 1};
    double scalesD = 0.0;
    double scalesH = 0.0;
    double scalesW = 0.0;
    // Create the gradOut aclTensor.
    ret = CreateAclTensor(gradOutHostData, gradOutShape, &gradOutDeviceAddr, aclDataType::ACL_FLOAT, &gradOut);
    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);

    const aclIntArray* outputSize = aclCreateIntArray(outputSizeData.data(), outputSizeData.size());
    CHECK_RET(outputSize != nullptr, return ACL_ERROR_INTERNAL_ERROR);

    const aclIntArray* inputSize = aclCreateIntArray(inputSizeData.data(), inputSizeData.size());
    CHECK_RET(inputSize != nullptr, return ACL_ERROR_INTERNAL_ERROR);

    // 3. Call the CANN operator library API. Replace with the specific API.
    uint64_t workspaceSize = 0;
    aclOpExecutor* executor;
    // Call the first-phase API of aclnnUpsampleNearestExact3dBackward.
    ret = aclnnUpsampleNearestExact3dBackwardGetWorkspaceSize(
        gradOut, outputSize, inputSize, scalesD, scalesH, scalesW, gradInput, &workspaceSize, &executor);
    CHECK_RET(ret == ACL_SUCCESS,
              LOG_PRINT("aclnnUpsampleNearestExact3dBackwardGetWorkspaceSize 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 aclnnUpsampleNearestExact3dBackward.
    ret = aclnnUpsampleNearestExact3dBackward(workspaceAddr, workspaceSize, executor, stream);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnUpsampleNearestExact3dBackward failed. ERROR: %d\n", ret);
              return ret);
    // 4. (Boilerplate) Synchronously 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 device-side memory to host-side memory. Modify based on the specific API definition.
    auto size = GetShapeSize(gradInputShape);
    std::vector<float> resultData(size, 0);
    ret = aclrtMemcpy(
        resultData.data(), resultData.size() * sizeof(resultData[0]), gradInputDeviceAddr, 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. Release aclTensor and aclScalar. Modify based on the specific API definition.
    aclDestroyTensor(gradOut);
    aclDestroyTensor(gradInput);

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