python (12.9k questions)

javascript (9.2k questions)

reactjs (4.7k questions)

java (4.2k questions)

java (4.2k questions)

c# (3.5k questions)

c# (3.5k questions)

html (3.3k questions)

Questions - avx

AVX512: How to convert first 8 bytes into 8 64-bit integers?

I have an __m512i inputVector where each of the 64 bytes contains some offset. Next I need to add the first 8 byte offsets to 8 64-bit values stored in another __m512i variable (base). (In order to pr...
test-img

Linoliumz

c++

c

avx

avx512

Votes: 0

Answers: 2

Latest Answer

_mm512_cvtepu8_epi64 does exactly what you want for the low 8 bytes. It only looks at the low 8 bytes of the input __m128i; you don't need to do anything special to make your 8 elements fill an __m12...
test-img

Peter Cordes

Does anyone know of a fix for an MSVC compiler bug/annoyance where SIMD Extension settings get "stuck" on AVX?

Does anyone know of a fix for an MSVC compiler bug/annoyance where SIMD Extension settings get "stuck" on AVX? The context of this question is coding up SIMD CPU dispatchers, closely followi...
test-img

dts

c++

visual-c++

simd

avx

vector-class-library

Votes: 0

Answers: 2

Latest Answer

If you want a single binary which works on SSE-only computers, but can leverage AVX when available, you need to do following. At the project level, set “Enable enhanced instruction set: Not set” if y...
test-img

Soonts

Efficient code to load AVX vectors for 1D convolution kernel of length 8

An implementation of a 1D convolution operation will often need to load a vectors of data that sequentially step through a buffer of data offset by one element each iteration. For example, consider a ...
test-img

TrentP

performance

optimization

simd

convolution

avx

Votes: 0

Answers: 2

Latest Answer

First of all, you should note that if your convolution is separable, this is very often worth doing. Simple example: res[i] = x[i]+x[i+1]+x[i+2]+x[i+3]+x[i+4]+x[i+5]+x[i+6]+x[i+7]; This can be done b...
test-img

chtz

Generate random numbers in a given range with AVX2, faster than SVML _mm256_rem_epu32 remainder?

I'm currently trying to implement an XOR_SHIFT Random Number Generator using AVX2, it's actually quite easy and very fast. However I need to be able to specify a range. This usually requires modulo. T...
test-img

dave_thenerd

c++

random

simd

modulo

avx

Votes: 0

Answers: 1

Latest Answer

If you range is smaller than ~16.7 million, and you don’t need cryptography-grade quality of the distribution, an easy and relatively fast method of narrowing these random numbers is FP32 math. Here’s...
test-img

Soonts

Posts

Questions

Blogs

Jobs