Re:Avoiding instruction cache misses (2019)
by jcelerier on Friday January 08, 2021 @08:57AM
(#25684740)
> Fast and slow are relative, that’s for sure. But assuming there is a problem before discovering it... isn’t that the definition of premature optimization?
that works if you have a definite performance target, say, "less than 5ms". Not if your performance target is, "it must be faster by a good margin than all the competitors that we know and don't know about, and stay that way"
that works if you have a definite performance target, say, "less than 5ms". Not if your performance target is, "it must be faster by a good margin than all the competitors that we know and don't know about, and stay that way"