EnQue

Applicability

Product	Supported
Atlas 350 Accelerator Card	√
Atlas A3 training product/Atlas A3 inference product	√
Atlas A2 training product/Atlas A2 inference product	√
Atlas 200I/500 A2 inference product	√
Atlas inference product AI Core	√
Atlas inference product Vector Core	x
Atlas training product	√

Function Usage

Pushes a tensor to a queue.

Prototype

The source and destination locations do not need to be specified.

template <typename T>
__aicore__ inline bool EnQue(const LocalTensor<T>& tensor)

The source and destination locations need to be specified.
Bind VECIN and VECOUT by using TQueBind to implement VECIN and VECOUT buffer reuse. The following API is used to implement buffer reuse for vector computation. The source and destination locations need to be specified during enqueuing. In the scenario where vector computation is not involved, call bool EnQue(LocalTensor<T>& tensor).
1 2
template <TPosition srcUserPos, TPosition dstUserPos, typename T> __aicore__ inline bool EnQue(const LocalTensor<T>& tensor)

Parameters

**Table 1** Template parameters
Parameter	Description
T	Data type of the tensor.
srcUserPos	src position of the queue. Supported path: GM->VECIN/VECOUT->GM
dstUserPos	dst position of the queue. Supported path: GM->VECIN/VECOUT->GM

**Table 2** Parameters
Parameter	Input/Output	Meaning
tensor	Input	Specified tensor.

Constraints

None

Returns

true: The tensor is successfully added to the queue.
false: The queue is full and the addition fails.

Example

// API: EnQue Tensor
AscendC::TPipe pipe;
AscendC::TQue<AscendC::TPosition::VECOUT, 4> que;
int num = 4;
int len = 1024;
pipe.InitBuffer(que, num, len);
AscendC::LocalTensor<half> tensor1 = que.AllocTensor<half>();
que.EnQue(tensor1);// Add the tensor to the queue of VECOUT.
// API: EnQue specifies a specific src/dst position to join the corresponding queue.
// template <TPosition srcUserPos, TPosition dstUserPos> bool EnQue(LocalTensor<T>& tensor)
AscendC::TPipe pipe;
AscendC::TQueBind<AscendC::TPosition::VECIN, AscendC::TPosition::VECOUT, 1> que;
int num = 4;
int len = 1024;
pipe.InitBuffer(que, num, len);
AscendC::LocalTensor<half> tensor1 = que.AllocTensor<half>();
que.EnQue<AscendC::TPosition::GM, AscendC::TPosition::VECIN, half>(tensor1);// Add tensors to the VECIN queue for buffer reuse.

Parent topic: TQue