aclnnGeGluV3Backward

📄 View Source Code

Applicable Products

Function

Description: Performs the backward propagation of aclnnGeGluV3.

Function Prototype

Each operator is divided into a two-phase API. First call the "aclnnGeGluV3BackwardGetWorkspaceSize" API to obtain the workspace size required for computation and the executor that contains the operator computation process, and then call the "aclnnGeGluV3Backward" API to perform the computation.

aclnnStatus aclnnGeGluV3BackwardGetWorkspaceSize(
  const aclTensor *gradOutput,
  const aclTensor *self,
  const aclTensor *gelu,
  int64_t          dim,
  int64_t          approximate,
  bool             activateLeft,
  aclTensor       *gradInput,
  uint64_t        *workspaceSize,
  aclOpExecutor  **executor)

aclnnStatus aclnnGeGluV3Backward(
  void            *workspace,
  uint64_t         workspaceSize,
  aclOpExecutor   *executor,
  aclrtStream      stream)

aclnnGeGluV3BackwardGetWorkspaceSize

• Parameters:

  Parameter	Input/Output	Description	Instruction	Data Type	Data Format	Dimension (Shape)	Non-contiguous Tensor
  gradOutput (aclTensor*)	Input	Input parameter for GeGluV3Backward computation, corresponding to gradOutput in the formula.	Except for the dim dimension, the sizes of other dimensions in shape are the same as self, and the size of the dim dimension is half that of self.	FLOAT, FLOAT16, BFLOAT16	ND	1-8	√
  self (aclTensor*)	Input	Input parameter of the GeGluV3Backward computation, which is self in the formula.	In shape, all dimensions except dim are the same size as gradOutput, and the dim dimension is twice the size of gradOutput	FLOAT, FLOAT16, BFLOAT16	ND	1-8	√
  gelu (aclTensor*)	Input	Input parameter for the GeGluV3Backward computation, corresponding to gelu in the formula.	The shape must be the same as gradOutput The data type must be the same as self.	FLOAT, FLOAT16, BFLOAT16	ND	1-8	√
  dim (int64_t)	Input	-	The value range is [-self.dim(), self.dim()-1].	-	-	-	-
  approximate (int64_t)	Input.	-	The value range is 0 ('none') and 1 ('tanh'). Atlas inference products only support 1 ('tanh').	-	-	-	-
  activateLeft (bool)	Input	Indicates the direction of the activation function operating on the data block.	false indicates that activation is applied to the right side.	-	-	-	-
  gradInput (aclTensor*)	Output	Output parameter of the GeGluV3Backward computation.	The data type is consistent with self. The shape must be the same as self.	FLOAT, FLOAT16, BFLOAT16	ND	1-8	√
  workspaceSize（uint64_t*）	Output	Returns the workspace size to be applied for on the Device side.	-	-	-	-	-
  executor (aclOpExecutor**)	Output	Returns the operator executor, which contains the operator computation flow.	-	-	-	-	-

  • Atlas inference products, Atlas training products: The supported data types are FLOAT32 and FLOAT16.

• 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 gradOutput, self, gelu, and gradInput parameters are null pointers.
  ACLNN_ERR_PARAM_INVALID	161002	The data types of the parameters gradOutput, self, gelu, and gradInput are not within the supported range.
  		The shapes of the parameters gradOutput, self, gelu, and gradInput do not meet the requirements.
  		When self.dim()=0, the value range of the parameter dim is not within [-1, 0]; when self.dim()>0, the value range of the parameter dim is not within [-self.dim(), self.dim()-1].
  		The value of the parameter approximate is not 0 or 1.

aclnnGeGluV3Backward

• Parameters:

  Parameter	Input/Output	Description
  workspace	Input	Specifies the workspace memory address applied for on the Device side.
  workspaceSize	Input	Specifies the workspace size applied for on the Device side, which is obtained through the first-phase API aclnnGeGluV3BackwardGetWorkspaceSize.
  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:
  • aclnnGeGluV3Backward 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_geglu_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: 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** selfOrResult)
{
    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);
    // Copy host data to the device memory by calling aclrtMemcpy.
    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 selfOrResult.
    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.
    *selfOrResult = 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);

    // Construct the input and output. Customize the construction based on the API.
    std::vector<int64_t> gradOutputShape = {2, 1};
    std::vector<int64_t> selfShape = {2, 2};
    std::vector<int64_t> geluShape = {2, 1};
    std::vector<int64_t> gradInputShape = {2, 2};
    void* gradOutputDeviceAddr = nullptr;
    void* selfDeviceAddr = nullptr;
    void* geluDeviceAddr = nullptr;
    void* gradInputDeviceAddr = nullptr;
    aclTensor* gradOutput = nullptr;
    aclTensor* self = nullptr;
    aclTensor* gelu = nullptr;
    aclTensor* gradInput = nullptr;
    std::vector<float> gradOutputHostData = {-2, -1};
    std::vector<float> selfHostData = {-2, -1, 0, 1};
    std::vector<float> geluHostData = {-2, -1};
    int64_t dim = -1;
    int64_t approximate = 1;
    std::vector<float> gradInputHostData = {0, 0, 0, 0};
    // Create the gradOutput aclTensor.
    ret = CreateAclTensor(
        gradOutputHostData, gradOutputShape, &gradOutputDeviceAddr, aclDataType::ACL_FLOAT16, &gradOutput);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    // Create the self aclTensor.
    ret = CreateAclTensor(selfHostData, selfShape, &selfDeviceAddr, aclDataType::ACL_FLOAT16, &self);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    // Create the gelu aclTensor.
    ret = CreateAclTensor(geluHostData, geluShape, &geluDeviceAddr, aclDataType::ACL_FLOAT16, &gelu);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    // Create the gradInput aclTensor.
    ret =
        CreateAclTensor(gradInputHostData, gradInputShape, &gradInputDeviceAddr, aclDataType::ACL_FLOAT16, &gradInput);
    CHECK_RET(ret == ACL_SUCCESS, return ret);

    // 3. Call the CANN operator library API. Replace with the actual API name.
    uint64_t workspaceSize = 0;
    aclOpExecutor* executor;
    bool activateLeft = false;
    // Call the first-phase API of aclnnGeGluV3Backward.
    ret = aclnnGeGluV3BackwardGetWorkspaceSize(
        gradOutput, self, gelu, dim, approximate, activateLeft, gradInput, &workspaceSize, &executor);
    CHECK_RET(
        ret == ACL_SUCCESS, LOG_PRINT("aclnnGeGluV3BackwardGetWorkspaceSize 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 aclnnGeGluV3Backward.
    ret = aclnnGeGluV3Backward(workspaceAddr, workspaceSize, executor, stream);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnGeGluV3Backward 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. 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(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 based on the specific API definition.
    aclDestroyTensor(gradOutput);
    aclDestroyTensor(self);
    aclDestroyTensor(gelu);
    aclDestroyTensor(gradInput);

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