Documentation

vaddreluconv

Function

vaddreluconv instruction abstraction.

z = vaddreluconv(x, y), where x and y are added by element, the relu value is calculated, and the output is quantized.

The following cast types are supported: FP16->INT8, FP32->FP16, and INT16->INT8.

Prototype

1
class vaddreluconv(x, y, z)

Parameters

Parameter

Input/Output

Data Type

Description

x

Input

Tensor variable

Input x-vector tensor. FP16, FP32, and INT16 are supported.

y

Input

Tensor variable

Input y-vector tensor. FP16, FP32, and INT16 are supported.

z

Output

Tensor variable

Output vector tensor. FP16 and INT8 are supported.

Example

1
2
3
4
5
6
from mskpp import vaddreluconv, Tensor
ub_x, ub_y, ub_z = Tensor("UB"), Tensor("UB"), Tensor("UB")
gm_x, gm_y = Tensor("GM"), Tensor("GM")
ub_x.load(gm_x)
ub_y.load(gm_y)
out = vaddreluconv(ub_x, ub_y, ub_z)()