如何将字节数组加载到__m128i？ [英] How to load byte array to __m128i ?

问题描述

大家好。



我将字节数组定义如下。



BYTE缓冲区[16] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15};



我打电话给'_mm_load_si128（）'。



__m128i current_0 = _mm_load_si128（（__ m128i *）缓冲区）;



但是，它发生错误消息'访问违规'。



这是错的吗？我真的不知道..



请给我建议。谢谢:)



我尝试了什么：



我尝试过如下代码。



BYTE缓冲区[16] = {0,1,2,3,4,5,6,7,8,9,10,11 ，12,13,14,15};

__m128i current_0 = _mm_loadu_si128（（__ m128i *）缓冲区）;



它有效。

但是我不知道当图像尺寸很大时，它们在性能上有'_mm_loadu_si128'和'_mm_load_si128'的差异。

Hi all.

I defined byte array as following.

BYTE buffer[16] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15};

And I call '_mm_load_si128()'.

__m128i current_0 = _mm_load_si128((__m128i*)buffer);

But, It occured error message 'Access Violation'.

Is it wrong? I really don't know..

Please give me advice. Thank you :)

What I have tried:

I tried as following code.

BYTE buffer[16] = {0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15};
__m128i current_0 = _mm_loadu_si128((__m128i*)buffer);

It worked.
But I don't know they have difference of performance '_mm_loadu_si128' and '_mm_load_si128' when the image has big size.

推荐答案

参见 _mm_loadu_si128 [ ^ ]。 _mm_load_si128 需要字节数组为16字节对齐。非对齐数组的性能会较慢。

See _mm_loadu_si128[^]. _mm_load_si128 needs the byte array to be 16-byte aligned. The performance will be slower for the non-aligned array.

这篇关于如何将字节数组加载到__m128i？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！