Skip to content

Comment on Beating the L1 cache with value speculation

Comments

So the solution is to write assembly for an extremely common micro-architecture design. Maybe the compilers should provide more tools to exploit such parallelism. That asm block looks like a weird hack.

AboutSource Built by g1lg1l

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