Skip to content

Comment on An Analysis of User-Space Idle State Instructions on x86 Processorsparent

Comments

Regarding your conclusion, I always figured that the reason WAITPKG seems kinda lame is the only reason they ported it to Core architectures was to make the heterogenous CPUs possible. It works better on Atom. On Core it does almost nothing, as you note. AMD's ISA extension was written from the ground up for their high performance server core, which might explain why it's actually useful.

Thanks for the history lesson, that makes a lot of sense.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.