News for hackers. Stuff that matters.
Re:Deep Learning for Compilers by varbhat on Tuesday September 01, 2020 @11:31AM (#24342879)
wouldn't this affect reproducible builds ?
Re:Deep Learning for Compilers by cec on Tuesday September 01, 2020 @05:44PM (#24347474)
Author here. In the sense that a trained neural net produces the same output from the same input, this is deterministic. But I don’t think that’s what you’re getting at. Where it gets interesting is if we inserted a feedback loop (such as FDO) so that the compiler could fine tune itself on past decisions. In that case, the build would still be deterministic, but the compiler would keep changing, invalidating the build cache.
Re:Deep Learning for Compilers by UncleMeat on Tuesday September 01, 2020 @01:11PM (#24344066)
Why?

Two of the projects (fuzzing, benchmark sets) don't affect the compiler behavior at all. The third project (heuristics) already interfere with reproducible builds if they use things like time budgets but could also be made entirely determistic.