This document is relevant for: Trn1, Trn2, Trn3

nki.language.trunc#

nki.language.trunc(x, dtype=None)[source]#

Truncated value of the input, element-wise.

((Similar to numpy.trunc))

Warning

This API is experimental and may change in future releases.

The truncated value of the scalar x is the nearest integer i which is closer to zero than x is. In short, the fractional part of the signed number x is discarded.

Parameters:

  • x – a tile.

  • dtype – (optional) data type to cast the output type to (see Supported Data Types for more information); if not specified, it will default to be the same as the data type of the input tile.

Returns:

a tile that has truncated values of x.

Examples:

import nki.language as nl

# nki.language.trunc -- truncates 3.7 toward zero to 3.0
a = nl.full((128, 512), 3.7, dtype=nl.float32, buffer=nl.sbuf)
c = nl.trunc(a)
expected = nl.full((128, 512), 3.0, dtype=nl.float32, buffer=nl.sbuf)
assert nl.equal(c, expected)

# nki.language.trunc -- truncates -3.7 toward zero to -3.0
a = nl.full((128, 512), -3.7, dtype=nl.float32, buffer=nl.sbuf)
c = nl.trunc(a)
expected = nl.full((128, 512), -3.0, dtype=nl.float32, buffer=nl.sbuf)
assert nl.equal(c, expected)

This document is relevant for: Trn1, Trn2, Trn3