aclnnAddLayerNorm

📄 View Source Code

Applicable Products

Function

• Description: Implements the AddLayerNorm operation.

• Formula:

  $$x = x1 + x2 + biasOptional$$ $$rstd = {{1}\over\sqrt {Var(x)+epsilon}}$$ $$y = (x-E(x)) * rstd * gamma + beta$$

  Where E(x) represents the mean and Var(x) represents the variance, both of which must be computed internally by the operator.

Function Prototype

Each operator uses a two-phase API. You must first call the aclnnAddLayerNormGetWorkspaceSize API to obtain the input parameters and calculate the required workspace size based on the computation process, and then call the aclnnAddLayerNorm API to perform the computation.

aclnnStatus aclnnAddLayerNormGetWorkspaceSize(
  const aclTensor *x1,
  const aclTensor *x2,
  const aclTensor *gamma,
  const aclTensor *beta,
  const aclTensor *biasOptional,
  double           epsilon,
  bool             additionalOutput,
  const aclTensor *yOut,
  const aclTensor *meanOut,
  const aclTensor *rstdOut,
  const aclTensor *xOut,
  uint64_t        *workspaceSize,
  aclOpExecutor  **executor)

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

aclnnAddLayerNormGetWorkspaceSize

• Parameters

  Parameter	Input/Output	Description	Instruction	Data Type	Data Format	Dimension (Shape)	Non-contiguous Tensor
  x1 (aclTensor*)	Input	Represents the input for the addition operation in AddLayerNorm. The operator performs the computation x1 + x2 + biasOptional and applies layer normalization to the result. Corresponds to `x1` in the formula.	Supports empty tensors. Does not support inputs where any dimension has a value of 0.	FLOAT32, FLOAT16, BFLOAT16	ND	1-8	√
  x2 (aclTensor*)	Input	Represents the input for the addition operation in AddLayerNorm. The operator performs x1 + x2 + biasOptional calculation and applies layer normalization to the result. Corresponds to `x2` in the formula.	Supports empty tensors. The shape must be consistent with `x1`.	FLOAT32, FLOAT16, BFLOAT16	ND	1-8	√
  beta (aclTensor*)	Input	Represents the beta parameter in layer normalization. Corresponds to `beta` in the corresponding formula.	Supports empty tensors. The dimension value of shape is the same as the dimension value of `x1` that requires normalization.	FLOAT32, FLOAT16, BFLOAT16	ND	1-8	√
  gamma (aclTensor*)	Input	Represents the gamma parameter in layer normalization. Corresponds to `gamma` in the corresponding formula.	Supports empty tensors. The dimension values of the shape must be the same as the dimension values of `x1` to be normalized.	FLOAT32, FLOAT16, BFLOAT16	ND	1-8	√
  biasOptional (aclTensor*)	Input	Optional input parameter, representing the input of the addition operation in AddLayerNorm. The operator performs the calculation x1 + x2 + biasOptional and applies layer normalization to the result. Corresponds to `biasOptional` in the formula.	Supports empty tensors. The shape can be consistent with `gamma`/`beta` or `x1`/`x2`.	FLOAT32, FLOAT16, BFLOAT16	ND	1-8	√
  epsilon (double)	Input	Specifies the value added to the denominator to ensure numerical stability. Corresponds to `epsilon` in the formula.	Only 1e-5 is supported.	-	-	-	-
  additionalOutput（bool）	Input	Indicates whether to enable the output of x = x1 + x2 + biasOptional.	-	-	-	-	-
  meanOut (aclTensor*)	Output	Represents the mean of the result of (x1 + x2 + biasOptional) during the LayerNorm computation. Corresponds to `E(x)` in the formula.	Supports empty tensors. The shape must satisfy the broadcast relationship with `x1` (the first several dimensions have the same dimensions as the first several dimensions of `x1`, the subsequent dimensions are 1, and the total number of dimensions is the same as that of `x1`; the first several dimensions refer to the dimensions of `x1` minus the dimensions of gamma, indicating the dimensions that do not require normalization). When the input `x1` is an empty tensor, `meanOut` must also be an empty tensor.	FLOAT32	ND	1-8	√
  rstdOut (aclTensor*)	Output	Indicates the result of `rstd` during the LayerNorm computation. Corresponds to `rstd` in the corresponding formula.	Supports empty tensors. The shape must satisfy the broadcast relationship with `x1` (the first several dimensions have the same dimensions as the first several dimensions of `x1`, the subsequent dimensions are 1, and the total number of dimensions is the same as that of `x1`. The first several dimensions refer to the dimensions of `x1` minus the dimensions of gamma, indicating the dimensions that do not require normalization). When the input `x1` is an empty tensor, `rstdOut` must also be an empty tensor.	FLOAT32	ND	1-8	√
  yOut (aclTensor*)	Output	Represents the output of the layer normalization result. Corresponds to `y` in the formula.	Supports empty tensors. The shape must be consistent with the input `x1`.	FLOAT32, FLOAT16, BFLOAT16	ND	1-8	√
  xOut (aclTensor*)	Output	Represents the output `x` of the Add result. Corresponds to `x` in the formula.	Supports empty tensors. The shape must be consistent with the input `x1`.	FLOAT32, FLOAT16, BFLOAT16	ND	1-8	√
  workspaceSize (uint64_t*)	Output	Returns the workspace size to be requested on the Device side.	-	-	-	-	-
  executor (aclOpExecutor**)	Output	Returns the op executor, which contains the operator computation flow.	-	-	-	-	-

  • Atlas inference products:

    • The data types of the parameters x1, x2, beta, gamma, biasOptional, yOut, and xOut do not support BFLOAT16.

    • The parameters meanOut and rstdOut are invalid in the current product usage scenario.

• 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	If the passed parameter is a required input, output, or required attribute and is a null pointer, 161001 is returned.
  ACLNN_ERR_PARAM_INVALID	161002	The data type of the input or output is not within the supported range.
  ACLNN_ERR_INNER_TILING_ERROR	561002	Fails to obtain the shapes of (x1, x2, gamma, beta, yOut, meanOut, rstdOut, xOut) during the tiling phase.
  		The dimension count of (x1, gamma) is greater than 8 or less than 0.
  		The dimension counts of (x1, x2, yOut, meanOut, rstdOut, xOut) are inconsistent.
  		The number of dimensions of x1 is less than that of gamma.
  		Any dimension of (x1, gamma, meanOut) is 0.
  		The shapes of (x1, x2, yOut, xOut) are not identical.
  		The shapes of (gamma, beta) are not identical.
  		The shapes of (meanOut, rstdOut) are not identical.
  		The dimensions of gamma differ from the dimensions of x to be normalized, or the dimensions of meanOut differ from the dimensions of x not to be normalized, or the dimensions of meanOut to be normalized are not 1.

aclnnAddLayerNorm

• 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 by the first-phase API aclnnAddLayerNormGetWorkspaceSize.
  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

• Functional Dimensions

  • Data Type Support

    • Atlas inference products: x1, x2, beta, gamma, and biasOptional support FLOAT32 and FLOAT16.

    • Atlas A2 training products/Atlas A2 inference products and Atlas A3 training products/Atlas A3 inference products: x1, x2, beta, gamma, and biasOptional support FLOAT32, FLOAT16, and BFLOAT16.

    • rstdOut and meanOut support: FLOAT32.

  • Data format support: ND.

  • Atlas inference products: The last axis length of the five inputs x1, x2, beta, gamma, and biasOptional must be greater than or equal to 32 bytes.

• Unsupported Type Description

  • DOUBLE: DOUBLE is not supported.

• Boundary Value Scenarios

  • When the input is Inf, the output is Inf.

  • When the input is NaN, the output is NaN.

• Supported Data Types by Product

  • Atlas A2 training products/Atlas A2 inference products, Atlas A3 training products/Atlas A3 inference products:

    x1 Data Type	x2 Data Type	gamma Data Type	beta Data Type	biasOptional Data Type	yOut Data Type	meanOut Data Type	rstdOut Data Type	xOut Data Type
    FLOAT32	FLOAT16	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32
    FLOAT32	BFLOAT16	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32
    FLOAT16	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32
    BFLOAT16	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32
    FLOAT16	FLOAT16	FLOAT32	FLOAT32	FLOAT16	FLOAT16	FLOAT32	FLOAT32	FLOAT16
    BFLOAT16	BFLOAT16	FLOAT32	FLOAT32	BFLOAT16	BFLOAT16	FLOAT32	FLOAT32	BFLOAT16
    FLOAT16	FLOAT16	FLOAT16	FLOAT16	FLOAT16	FLOAT16	FLOAT32	FLOAT32	FLOAT16
    BFLOAT16	BFLOAT16	BFLOAT16	BFLOAT16	BFLOAT16	BFLOAT16	FLOAT32	FLOAT32	BFLOAT16
    FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32

  • Atlas inference products:

    x1 Data Type	x2 Data Type	gamma Data Type	beta Data Type	biasOptional Data Type	yOut Data Type	meanOut Data Type	rstdOut Data Type	xOut Data Type
    FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32	FLOAT32
    FLOAT16	FLOAT16	FLOAT16	FLOAT16	FLOAT16	FLOAT16	FLOAT32	FLOAT32	FLOAT16

• Deterministic computation:

  • aclnnAddLayerNorm 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_add_layer_norm.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** 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 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. For details, 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 == ACL_SUCCESS, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);

    // 2.
    // Construct the inputs and outputs. Customize the construction based on the API interface. This example calls the API once without the optional biasOptional input and once with the biasOptional input.
    float eps = 1e-6;
    bool additionalOutput = true;

    std::vector<int64_t> x1Shape = {1, 2, 8};
    std::vector<int64_t> x2Shape = {1, 2, 8};
    std::vector<int64_t> gammaShape = {8};
    std::vector<int64_t> betaShape = {8};
    std::vector<int64_t> biasOptionalShape = {8};

    std::vector<int64_t> outputYShape = {1, 2, 8};
    std::vector<int64_t> outputMeanShape = {1, 2, 1};
    std::vector<int64_t> outputRstdShape = {1, 2, 1};
    std::vector<int64_t> outputXShape = {1, 2, 8};

    void* x1DeviceAddr = nullptr;
    void* x2DeviceAddr = nullptr;
    void* betaDeviceAddr = nullptr;
    void* gammaDeviceAddr = nullptr;
    void* biasOptionalDeviceAddr = nullptr;

    // Output Device address without biasOptional.
    void* outputYDeviceAddr = nullptr;
    void* outputMeanDeviceAddr = nullptr;
    void* outputRstdDeviceAddr = nullptr;
    void* outputXDeviceAddr = nullptr;

    // Output Device address with biasOptional.
    void* outputYDeviceAddrbiasOptional = nullptr;
    void* outputMeanDeviceAddrbiasOptional = nullptr;
    void* outputRstdDeviceAddrbiasOptional = nullptr;
    void* outputXDeviceAddrbiasOptional = nullptr;

    aclTensor* x1 = nullptr;
    aclTensor* x2 = nullptr;
    aclTensor* beta = nullptr;
    aclTensor* gamma = nullptr;
    aclTensor* biasOptional = nullptr;

    // Used for aclTensor without biasOptional.
    aclTensor* outputY = nullptr;
    aclTensor* outputMean = nullptr;
    aclTensor* outputRstd = nullptr;
    aclTensor* outputX = nullptr;

    // Used for aclTensor with biasOptional.
    aclTensor* outputYbiasOptional = nullptr;
    aclTensor* outputMeanbiasOptional = nullptr;
    aclTensor* outputRstdbiasOptional = nullptr;
    aclTensor* outputXbiasOptional = nullptr;

    std::vector<float> x1HostData = {1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2};
    std::vector<float> x2HostData = {4, 4, 4, 4, 4, 4, 4, 4, -3, -3, -3, -3, -3, -3, -3, -3};
    std::vector<float> gammaHostData = {2, 2, 2, 2, 2, 2, 2, 2};
    std::vector<float> betaHostData = {0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1};
    std::vector<float> biasOptionalHostData = {0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5};

    // Used for HostData without biasOptional.
    std::vector<float> outputYHostData(1 * 2 * 8);
    std::vector<float> outputMeanHostData(2);
    std::vector<float> outputRstdHostData(2);
    std::vector<float> outputXHostData(1 * 2 * 8);

    // Used for HostData with biasOptional.
    std::vector<float> outputYHostDatabiasOptional(1 * 2 * 8);
    std::vector<float> outputMeanHostDatabiasOptional(2);
    std::vector<float> outputRstdHostDatabiasOptional(2);
    std::vector<float> outputXHostDatabiasOptional(1 * 2 * 8);

    // Create the self aclTensor.
    ret = CreateAclTensor(x1HostData, x1Shape, &x1DeviceAddr, aclDataType::ACL_FLOAT, &x1);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(x2HostData, x2Shape, &x2DeviceAddr, aclDataType::ACL_FLOAT, &x2);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(betaHostData, betaShape, &betaDeviceAddr, aclDataType::ACL_FLOAT, &beta);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(gammaHostData, gammaShape, &gammaDeviceAddr, aclDataType::ACL_FLOAT, &gamma);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(
        biasOptionalHostData, biasOptionalShape, &biasOptionalDeviceAddr, aclDataType::ACL_FLOAT, &biasOptional);
    CHECK_RET(ret == ACL_SUCCESS, return ret);

    // Create an aclTensor without biasOptional.
    ret = CreateAclTensor(outputYHostData, outputYShape, &outputYDeviceAddr, aclDataType::ACL_FLOAT, &outputY);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(
        outputMeanHostData, outputMeanShape, &outputMeanDeviceAddr, aclDataType::ACL_FLOAT, &outputMean);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(
        outputRstdHostData, outputRstdShape, &outputRstdDeviceAddr, aclDataType::ACL_FLOAT, &outputRstd);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(outputXHostData, outputXShape, &outputXDeviceAddr, aclDataType::ACL_FLOAT, &outputX);
    CHECK_RET(ret == ACL_SUCCESS, return ret);

    // Create an aclTensor with biasOptional.
    ret = CreateAclTensor(
        outputYHostDatabiasOptional, outputYShape, &outputYDeviceAddrbiasOptional, aclDataType::ACL_FLOAT,
        &outputYbiasOptional);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(
        outputMeanHostDatabiasOptional, outputMeanShape, &outputMeanDeviceAddrbiasOptional, aclDataType::ACL_FLOAT,
        &outputMeanbiasOptional);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(
        outputRstdHostDatabiasOptional, outputRstdShape, &outputRstdDeviceAddrbiasOptional, aclDataType::ACL_FLOAT,
        &outputRstdbiasOptional);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(
        outputXHostDatabiasOptional, outputXShape, &outputXDeviceAddrbiasOptional, aclDataType::ACL_FLOAT,
        &outputXbiasOptional);
    CHECK_RET(ret == ACL_SUCCESS, return ret);

    // Example of calling the aclnnAddLayerNorm API, including one call with biasOptional and one without.
    // 3. Call the CANN operator library API. Replace with the specific API name.

    // 3.1 Example without the biasOptional optional input.
    // Call the first-phase API of aclnnAddLayerNorm.
    uint64_t workspaceSize = 0;
    aclOpExecutor* executor;
    LOG_PRINT("\nUse aclnnAddLayerNorm Non-biasOptional Port.");
    // You can directly pass nullptr for the biasOptional parameter.
    ret = aclnnAddLayerNormGetWorkspaceSize(
        x1, x2, gamma, beta, nullptr, eps, additionalOutput, outputY, outputMean, outputRstd, outputX, &workspaceSize,
        &executor);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnAddLayerNormGetWorkspaceSize 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 aclnnAddLayerNorm.
    ret = aclnnAddLayerNorm(workspaceAddr, workspaceSize, executor, stream);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnAddLayerNorm failed. ERROR: %d\n", ret); return ret);

    // 3.2 Example with the biasOptional input.
    // Call the first-phase API of aclnnAddLayerNorm.
    uint64_t workspaceSizebiasOptional = 0;
    aclOpExecutor* executorbiasOptional;
    LOG_PRINT("\nUse aclnnAddLayerNorm biasOptional Port.");
    // Pass biasOptional normally.
    ret = aclnnAddLayerNormGetWorkspaceSize(
        x1, x2, gamma, beta, biasOptional, eps, additionalOutput, outputYbiasOptional, outputMeanbiasOptional,
        outputRstdbiasOptional, outputXbiasOptional, &workspaceSizebiasOptional, &executorbiasOptional);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnAddLayerNormGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);
    // Apply for device memory based on the workspaceSize calculated by the first-phase API.
    void* workspaceAddrbiasOptional = nullptr;
    if (workspaceSizebiasOptional > 0) {
        ret = aclrtMalloc(&workspaceAddrbiasOptional, workspaceSizebiasOptional, 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 aclnnAddLayerNorm.
    ret = aclnnAddLayerNorm(workspaceAddrbiasOptional, workspaceSizebiasOptional, executorbiasOptional, stream);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnAddLayerNorm 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.

    // 5.1 Copy the output without biasOptional.
    auto outputYSize = GetShapeSize(outputYShape);
    std::vector<float> resultDataY(outputYSize, 0);
    ret = aclrtMemcpy(
        resultDataY.data(), resultDataY.size() * sizeof(resultDataY[0]), outputYDeviceAddr,
        outputYSize * sizeof(resultDataY[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);
    LOG_PRINT("==== AddLayerNorm non-biasOptional: y output");
    for (int64_t i = 0; i < outputYSize; i++) {
        LOG_PRINT("result[%ld] is: %f\n", i, resultDataY[i]);
    }

    auto outputMeanSize = GetShapeSize(outputMeanShape);
    std::vector<float> resultDataMean(outputMeanSize, 0);
    ret = aclrtMemcpy(
        resultDataMean.data(), resultDataMean.size() * sizeof(resultDataMean[0]), outputMeanDeviceAddr,
        outputMeanSize * sizeof(resultDataMean[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);
    LOG_PRINT("==== AddLayerNorm non-biasOptional: mean output");
    for (int64_t i = 0; i < outputMeanSize; i++) {
        LOG_PRINT("result[%ld] is: %f\n", i, resultDataMean[i]);
    }

    auto outputRstdSize = GetShapeSize(outputRstdShape);
    std::vector<float> resultDataRstd(outputRstdSize, 0);
    ret = aclrtMemcpy(
        resultDataRstd.data(), resultDataRstd.size() * sizeof(resultDataRstd[0]), outputRstdDeviceAddr,
        outputRstdSize * sizeof(resultDataRstd[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);
    LOG_PRINT("==== AddLayerNorm non-biasOptional: rstd output");
    for (int64_t i = 0; i < outputRstdSize; i++) {
        LOG_PRINT("result[%ld] is: %f\n", i, resultDataRstd[i]);
    }

    auto outputXSize = GetShapeSize(outputXShape);
    std::vector<float> resultDataX(outputXSize, 0);
    ret = aclrtMemcpy(
        resultDataX.data(), resultDataX.size() * sizeof(resultDataX[0]), outputXDeviceAddr,
        outputXSize * sizeof(resultDataX[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);
    LOG_PRINT("==== AddLayerNorm non-biasOptional: x output");
    for (int64_t i = 0; i < outputXSize; i++) {
        LOG_PRINT("result[%ld] is: %f\n", i, resultDataX[i]);
    }

    // 5.2 Copy the output with biasOptional.
    auto outputYSizebiasOptional = GetShapeSize(outputYShape);
    std::vector<float> resultDataYbiasOptional(outputYSizebiasOptional, 0);
    ret = aclrtMemcpy(
        resultDataYbiasOptional.data(), resultDataYbiasOptional.size() * sizeof(resultDataYbiasOptional[0]),
        outputYDeviceAddrbiasOptional, outputYSizebiasOptional * sizeof(resultDataYbiasOptional[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);
    LOG_PRINT("==== AddLayerNorm biasOptional: y output");
    for (int64_t i = 0; i < outputYSizebiasOptional; i++) {
        LOG_PRINT("result[%ld] is: %f\n", i, resultDataYbiasOptional[i]);
    }

    auto outputMeanSizebiasOptional = GetShapeSize(outputMeanShape);
    std::vector<float> resultDataMeanbiasOptional(outputMeanSizebiasOptional, 0);
    ret = aclrtMemcpy(
        resultDataMeanbiasOptional.data(), resultDataMeanbiasOptional.size() * sizeof(resultDataMeanbiasOptional[0]),
        outputMeanDeviceAddrbiasOptional, outputMeanSizebiasOptional * sizeof(resultDataMeanbiasOptional[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);
    LOG_PRINT("==== AddLayerNorm biasOptional: mean output");
    for (int64_t i = 0; i < outputMeanSizebiasOptional; i++) {
        LOG_PRINT("result[%ld] is: %f\n", i, resultDataMeanbiasOptional[i]);
    }

    auto outputRstdSizebiasOptional = GetShapeSize(outputRstdShape);
    std::vector<float> resultDataRstdbiasOptional(outputRstdSizebiasOptional, 0);
    ret = aclrtMemcpy(
        resultDataRstdbiasOptional.data(), resultDataRstdbiasOptional.size() * sizeof(resultDataRstdbiasOptional[0]),
        outputRstdDeviceAddrbiasOptional, outputRstdSizebiasOptional * sizeof(resultDataRstdbiasOptional[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);
    LOG_PRINT("==== AddLayerNorm biasOptional: rstd output");
    for (int64_t i = 0; i < outputRstdSizebiasOptional; i++) {
        LOG_PRINT("result[%ld] is: %f\n", i, resultDataRstdbiasOptional[i]);
    }

    auto outputXSizebiasOptional = GetShapeSize(outputXShape);
    std::vector<float> resultDataXbiasOptional(outputXSizebiasOptional, 0);
    ret = aclrtMemcpy(
        resultDataXbiasOptional.data(), resultDataXbiasOptional.size() * sizeof(resultDataXbiasOptional[0]),
        outputXDeviceAddrbiasOptional, outputXSizebiasOptional * sizeof(resultDataXbiasOptional[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);
    LOG_PRINT("==== AddLayerNorm biasOptional: x output");
    for (int64_t i = 0; i < outputXSizebiasOptional; i++) {
        LOG_PRINT("result[%ld] is: %f\n", i, resultDataXbiasOptional[i]);
    }

    // 6. Release aclTensor and aclScalar. Modify based on the specific API definition.
    aclDestroyTensor(x1);
    aclDestroyTensor(x2);
    aclDestroyTensor(beta);
    aclDestroyTensor(gamma);
    aclDestroyTensor(biasOptional);

    aclDestroyTensor(outputY);
    aclDestroyTensor(outputMean);
    aclDestroyTensor(outputRstd);
    aclDestroyTensor(outputX);

    aclDestroyTensor(outputYbiasOptional);
    aclDestroyTensor(outputMeanbiasOptional);
    aclDestroyTensor(outputRstdbiasOptional);
    aclDestroyTensor(outputXbiasOptional);

    // 7. Release device resources. Modify based on the specific API definition.
    aclrtFree(x1DeviceAddr);
    aclrtFree(x2DeviceAddr);
    aclrtFree(gammaDeviceAddr);
    aclrtFree(betaDeviceAddr);
    aclrtFree(biasOptionalDeviceAddr);

    aclrtFree(outputYDeviceAddr);
    aclrtFree(outputMeanDeviceAddr);
    aclrtFree(outputRstdDeviceAddr);
    aclrtFree(outputXDeviceAddr);

    aclrtFree(outputYDeviceAddrbiasOptional);
    aclrtFree(outputMeanDeviceAddrbiasOptional);
    aclrtFree(outputRstdDeviceAddrbiasOptional);
    aclrtFree(outputXDeviceAddrbiasOptional);

    if (workspaceSize > 0) {
        aclrtFree(workspaceAddr);
    }

    if (workspaceSizebiasOptional > 0) {
        aclrtFree(workspaceAddrbiasOptional);
    }

    aclrtDestroyStream(stream);
    aclrtResetDevice(deviceId);
    aclFinalize();
    return 0;
}