@SuperDicq @emilis You'd probably need to mess with microcode and it would kill the performance of your machine
@emilis@jigglypuff.club @valleyforge@qoto.org idk, just run some other tasks on the cpu so it forgets the data?
@SuperDicq@minidisc.tokyo @valleyforge@qoto.org great thing when you’re trying to benchmark software :thumbsuphelp:
@emilis@jigglypuff.club @valleyforge@qoto.org Also just upload your benchmarks with branch prediction, it will look good to clients.
@emilis@jigglypuff.club @valleyforge@qoto.org Alternatively, just run it on a new machine.
@SuperDicq@minidisc.tokyo @valleyforge@qoto.org I wanna run like 1m iterations tho
@SuperDicq@minidisc.tokyo @valleyforge@qoto.org I’m not trying to sell this, this is more “is this sufficient or should I optimize” thing for my own interest
@valleyforge@qoto.org @SuperDicq@minidisc.tokyo yeah I don’t want this as a permanent thing, just something to get more realistic benchmarks without having to deal with diverse benchmark data (so it doesn’t just hit branch prediction correctly every damn time)