aclnnGeGluV3

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: Gaussian error linear unit activation gating function. Compared with aclnnGeGlu, this API extends the capability to set the direction of the data block for the activation function operation.
Formula:

If activateLeft is true, the activation is applied to the left half of $self$ .
$out_{i}=GeGlu(self_{i}) = Gelu(A) \cdot B$
If activateLeft is false, it indicates that the right half of $self$ is activated.
$out_{i}=GeGlu(self_{i}) = A \cdot Gelu(B)$
Where $A$ represents the left half of $self$ , and $B$ represents the right half of $self$ .

Function Prototype

Each operator is divided into a two-phase API. The "aclnnGeGluV3GetWorkspaceSize" API must be called first to obtain the workspace size required for computation and the executor that encapsulates the operator computation flow. Then, the "aclnnGeGluV3" API is called to perform the computation.

aclnnStatus aclnnGeGluV3GetWorkspaceSize(
  const aclTensor   *self,
  int64_t            dim,
  int64_t            approximate,
  bool               activateLeft,
  aclTensor         *out,
  aclTensor         *outGelu,
  uint64_t          *workspaceSize,
  aclOpExecutor    **executor)

aclnnStatus aclnnGeGluV3(
  void              *workspace,
  uint64_t           workspaceSize,
  aclOpExecutor     *executor,
  aclrtStream        stream)

aclnnGeGluV3GetWorkspaceSize

Parameters:

Parameter	Input/Output	Description	Instruction	Data Type	Data Format	Dimension (Shape)	Non-contiguous Tensor
self (aclTensor*)	Input	The input parameter for GeGluV3 computation, which is "self" in the formula.	-	FLOAT、FLOAT16、BFLOAT16	ND	0-8	√
dim (int64_t)	Input	Optional input parameter.	The specified slice axis must split the corresponding axis of self into two halves. The axis of self corresponding to dim must be an even number.	-	-	-	-
approximate (int64_t)	Input	Optional input parameter.	Index of the activation function used in GeGluV3 computation. 0 indicates none, and 1 indicates tanh.	-	-	-	-
activateLeft (bool)	Input	Indicates the direction of the activation function operation on the data block.	false indicates activation on the right side, and true indicates activation on the left side.	-	-	-	-
out (aclTensor*)	Output	Output parameter of the GeGluV3 computation, i.e., out_i in the formula.	The shape of out must be consistent with the shape of self except for the axis specified by dim, where the axis specified by dim is half the size of the corresponding axis of self. The data type is the same as that of self.	FLOAT, FLOAT16, BFLOAT16	ND	0-8	√
outGelu (aclTensor*)	Output	Output parameter of the GeGluV3 computation.	The shape of outGelu must be consistent with the shape of self except for the axis specified by dim, where the axis specified by dim is half of the corresponding axis of the shape of self. The data type is the same as that of self.	FLOAT, FLOAT16, BFLOAT16	ND	0-8	√
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, Atlas training products: The supported data types are FLOAT 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	Indicates that the self, out, or outGelu parameter is a null pointer.
ACLNN_ERR_PARAM_INVALID	161002	The data types of the parameters self, out, and outGelu are not within the supported range.
		The data types of the parameters out and outGelu are inconsistent with that of self.
		The number of dimensions of self, out, or outGelu is greater than 8.
		When self.dim()=0, the value of dim is not within the range of [-1, 0]; when self.dim()>0, the value of dim is not within the range of [-self.dim, self.dim()-1].
		The shapes of out and outGelu must be consistent with the shape of self except for the axis specified by dim, where the axis specified by dim is half of the corresponding axis of self.

aclnnGeGluV3

Parameters:

Parameter	Input/Output	Description
workspace	Input	Memory address of the workspace applied on the Device side.
workspaceSize	Input	Size of the workspace applied on the Device side, obtained from the first-phase API aclnnGeGluV3GetWorkspaceSize.
executor	Input	Operator executor that contains the operator computation flow.
stream	Input	The stream that executes the task.

Return value

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

Constraints

Deterministic computation:
- aclnnGeGluV3 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.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);
    // Copy data from the host to the device using 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 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];
    }

    // Create an aclTensor by calling aclCreateTensor.
    *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.
    // Enter the actual device ID.
    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 interface.
    std::vector<int64_t> selfShape = {2, 2};
    std::vector<int64_t> outShape = {2, 1};
    void* selfDeviceAddr = nullptr;
    void* outDeviceAddr = nullptr;
    void* outGeluDeviceAddr = nullptr;
    aclTensor* self = nullptr;
    aclTensor* out = nullptr;
    aclTensor* outGelu = nullptr;
    std::vector<float> selfHostData = {0, 1, 2, 3};
    std::vector<float> outHostData = {0, 0};
    std::vector<float> outGeluHostData = {0, 0};
    int dim = -1;
    int approximate = 1;
    // Create the self aclTensor.
    ret = CreateAclTensor(selfHostData, selfShape, &selfDeviceAddr, aclDataType::ACL_FLOAT, &self);
    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);
    // Create the outGelu aclTensor.
    ret = CreateAclTensor(outGeluHostData, outShape, &outGeluDeviceAddr, aclDataType::ACL_FLOAT, &outGelu);
    CHECK_RET(ret == ACL_SUCCESS, return ret);

    // 3. Call the CANN operator library API. Replace with the specific API name.
    uint64_t workspaceSize = 0;
    aclOpExecutor* executor;
    // Call the first-phase API of aclnnGeGluV3.
    bool activateLeft = false;
    ret = aclnnGeGluV3GetWorkspaceSize(self, dim, approximate, activateLeft, out, outGelu, &workspaceSize, &executor);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnGeGluV3GetWorkspaceSize 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 aclnnGeGluV3.
    ret = aclnnGeGluV3(workspaceAddr, workspaceSize, executor, stream);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnGeGluV3 failed. ERROR: %d\n", ret); return ret);

    // 4. (Boilerplate) Synchronously wait for the 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> resultData(size, 0);
    ret = aclrtMemcpy(
        resultData.data(), resultData.size() * sizeof(resultData[0]), outDeviceAddr, 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]);
    }

    std::vector<float> resultGeluData(size, 0);
    ret = aclrtMemcpy(
        resultGeluData.data(), resultGeluData.size() * sizeof(resultGeluData[0]), outGeluDeviceAddr,
        size * sizeof(resultGeluData[0]), ACL_MEMCPY_DEVICE_TO_HOST);
    CHECK_RET(
        ret == ACL_SUCCESS, LOG_PRINT("copy resultGelu 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, resultGeluData[i]);
    }

    // 6. Release the aclTensor and aclScalar. Modify based on the specific API definition.
    aclDestroyTensor(self);
    aclDestroyTensor(out);
    aclDestroyTensor(outGelu);

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