Hacker News Reader

Big fan of your site. Travis Downs has done some work on characterizing hardware transition latency. I believe his tools are on GitHub. https://travisdowns.github.io/blog/2020/01/17/avxfreq1.html

View on HN

Yeah, he has some pretty interesting stuff.

But AVX512 transition latency is pretty different from clocking up from idle. The CPU is already running at full tilt, and is making a relatively small frequency/voltage change. You can see from his writeup that it's well under a millisecond. I believe newer Intel CPUs have done away with AVX-512 downclocking completely.

I probably won't be able to look into that specific issue since I don't have any AVX-512 capable CPUs.

Yeah I think his insights still apply, though. The difference between a FIVR part and other parts might be significant, and since the CPU has to wait for the VR to settle it might not be appropriate to ascribe state transition latency entirely to the CPU generation. It’s really the whole platform.

Reply on news.ycombinator.com