News for hackers. Stuff that matters.

A comment on AVX-512, what’s useful for us. Parent comment

Re:AVX-512, what’s useful for us by Dylan16807 on Friday December 22, 2017 @01:27PM (#15989682)
What models don't slow down? Are you claiming the slowdown isn't actually necessary to keep the chip stable?
[ Parent | on HN ]
Re:AVX-512, what’s useful for us by justincormack on Friday December 22, 2017 @03:28PM (#15990633)
compare these

https://en.wikichip.org/wiki/intel/xeon_silver/4116#Frequenc...

https://en.wikichip.org/wiki/intel/xeon_gold/6154#Frequencie...

https://en.wikichip.org/wiki/intel/xeon_platinum/8180m#Frequ...

So platinum can run avx512 on 28 cores at 2.3GHz, while silver runs 12 cores at 1.4GHz

[ Parent | on HN ]
Re:AVX-512, what’s useful for us by brigade on Friday December 22, 2017 @05:00PM (#15991258)
From your links, gold/plat has a ~15% clock speed hit when using ymm muls, then another ~20% hit when going to zmm muls.

So for 512-bit wide to not slow things down on Intel's chips, you need ~30% runtime of all cores to already be in AVX.

[ Parent | on HN ]
Re:AVX-512, what’s useful for us by jabl on Friday December 22, 2017 @03:38PM (#15990711)
Xeon Phi, maybe? Since the entire raison d'être of that chip is doing avx-512 computation. Or if you're a glass half empty kind of person, you'd argue that it doesn't speed up when running scalar code. :)
[ Parent | on HN ]