SSE 比较返回 NAN 的向量 [英] SSE comparison returns vector of NANs

问题描述

我正在尝试这样的事情:

I am trying something like this:

__m128 cA = _mm_set_ps1(-2.0f);
__m128 cB = _mm_set_ps1(2.0f);
__m128 df = _mm_cmpgt_ps(cA, cB);

在这种情况下 df 返回零.

In this case df returns with zeros.

但如果我这样做:

__m128 cA = _mm_set_ps1(2.0f);
__m128 cB = _mm_set_ps1(-2.0f);
__m128 df = _mm_cmpgt_ps(cA, cB);

它返回所有 -nan .这是预期的行为吗?如果是，我如何评估这些nans?

It returns all -nan .Is is expected behaviour? If it is,how do I evaluate those nans?

使用英特尔 CPU，MS VisualStudio 2017

Using Intel CPU,MS VisualStudio 2017

推荐答案

SIMD 比较产生掩码.全 1 位是 -NaN 的位模式.全零位是 +0.0

SIMD compares produce a mask. All-one bits is the bit-pattern for -NaN. All-zero bits is the bit-pattern for +0.0

它们不应被解释为 float.将它们与 _mm_movemask_ps、混合或诸如 _mm_and_ps 之类的东西一起使用.

They're not intended to be interpreted as float. Use them with _mm_movemask_ps, blends, or things like _mm_and_ps.

例如_mm_and_ps(vec, cmp_result) 将比较为假的元素归零.您可以使用它通过在添加之前将某些输入元素归零来执行条件添加.

e.g. _mm_and_ps( vec, cmp_result) zeros the elements where the compare was false. You can use this to do a conditional add by zeroing some input elements before an add.

有关如何使用 SIMD 的更多信息，请查阅指南/教程.https://stackoverflow.com/tags/sse/info

For more about how to use SIMD, look up a guide/tutorial. https://stackoverflow.com/tags/sse/info

这篇关于SSE 比较返回 NAN 的向量的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！