Why are processors with only AVX superior to AVX2 processors for many SIMD algorithms?

Question

I study the benefits of SIMD algorithms in C # and C ++ and find that in many cases using 128-bit registers on an AVX processor provides a better improvement than using 256-bit registers on an AVX2 processor, but I don’t understand why.

By improvement, I mean accelerating the SIMD algorithm relative to the non-SIMD algorithm on the same machine.

Answer 1

AVX 256- , AVX ( , VEX). AVX, CPU FP - 70 , AVX 128 .

AVX 700 , .

, (doh!), (double doh!). , , AVX. , , " ", , AVX , , " ". , , , 2,3 , 2,7, AVX, 2,1 2,3, 1,9 (. 2-4 this).

10-15%, , 20% SLOWER, AVX. SIMD, 256- . AVX, "" , AVX, , , AVX.

, SIMD 128 256 , SIMD-. ... (, , , , , , ).

Answer 2

( )

, SIMD . , , , . AVX AVX2 , SIMD.