GetVecLen

Applicability

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

Function Usage

Obtains the vector length (VL) of RegTensor.

Prototype

__aicore__ inline constexpr uint32_t GetVecLen()

Parameters

None

Returns

Vector length, in bytes.

Restrictions

None

Example

The following example shows how to obtain the number of iterations using GetVecLen:

template <typename T>
__aicore__ inline void AddCustomImpl(__local_mem__ T *dst, __local_mem__ T *src0, __local_mem__ T *src1,
    uint32_t calCount)
{
    AscendC::Reg::RegTensor<T> reg0;
    AscendC::Reg::RegTensor<T> reg1;
    AscendC::Reg::RegTensor<T> reg2;
    AscendC::Reg::MaskReg mask;
    constexpr uint32_t repeatElm = AscendC::GetVecLen() / sizeof(T);
    uint16_t repeatTime = AscendC::CeilDivision(calCount, repeatElm);
    for (uint16_t i = 0; i < repeatTime; ++i) {
        mask = AscendC::Reg::UpdateMask<T>(calCount);
        AscendC::Reg::LoadAlign(reg0, src0 + i * repeatElm);
        AscendC::Reg::LoadAlign(reg1, src1 + i * repeatElm);
        AscendC::Reg::Add(reg2, reg0, reg1, mask);
        AscendC::Reg::StoreAlign(dst + i * repeatElm, reg2, mask);
    }
}

Parent topic: Tool Functions