Documentation

vrsqrt

Function

vrsqrt instruction abstraction.

Calculates the reciprocal square root of a floating point number.

Prototype

1
class vrsqrt(x, y)

Parameters

Parameter

Input/Output

Data Type

Description

x

Input

Tensor variable

Input x-vector tensor. float16 and float32 are supported.

y

Output

Tensor variable

Output vector tensor. float16 and float32 are supported.

Example

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