DeQue

Product Support

Product	Supported
Atlas A3 training products / Atlas A3 inference products	√
Atlas A2 training products / Atlas A2 inference products	√
Atlas 200I/500 A2 inference products	√
Atlas inference product 's AI Core	√
Atlas inference product 's Vector Core	x
Atlas training products	√

Function

Extracts tensors from the queue for subsequent processing.

Prototype

Non-inplace API: Obtains the enqueued LocalTensor address from the queue, assigns the address to the newly created tensor, and returns the address.

        
             template <typename T>
__aicore__ inline LocalTensor<T> DeQue()

Inplace API: Returns the result through output parameters, which reduces the overhead of repeatedly creating tensors. For details, see How to Improve Operator Performance Through Inplace Tensor Operations.

        
             template <typename T>
__aicore__ inline void DeQue(LocalTensor<T>& tensor)

Parameters

**Table 1** Template parameters
Parameter	Description
T	Data type of the tensor.

**Table 2** Parameters
Parameter	Input/Output	Meaning
tensor	Output	The inplace API returns the tensor through output parameters.

Restrictions

Performing DeQue on an empty queue is abnormal behavior, and an error is reported during CPU debugging.
For the non-inplace API, the depth template parameter of TQueBind must be set to a non-zero value. For the inplace API, the depth template parameter of TQueBind must be set to 0.

Returns

The return value of the non-inplace API is the LocalTensor obtained from the queue. The inplace API does not return any value.

Example

Non-inplace API

        
             AscendC::TPipe pipe;
AscendC::TQueBind<AscendC::TPosition::VECOUT, AscendC::TPosition::GM, 4> que;
int num = 4;
int len = 1024;
pipe.InitBuffer(que, num, len);
AscendC::LocalTensor<half> tensor1 = que.AllocTensor<half>();
que.EnQue(tensor1);
AscendC::LocalTensor<half> tensor2 = que.DeQue<half>(); // Move the tensor out of the VECOUT queue.

Inplace API

        
             AscendC::TPipe pipe;
AscendC::TQueBind<AscendC::TPosition::VECOUT, AscendC::TPosition::GM, 0> que;
int num = 2;
int len = 1024;
pipe.InitBuffer(que, num, len);
AscendC::LocalTensor<half> tensor1;
que.AllocTensor<half>(tensor1);
que.EnQue(tensor1);
que.DeQue<half>(tensor1); // Move the tensor out of the VECOUT queue.
que.FreeTensor<half>(tensor1);

Parent topic: TQueBind