deeply-nested-opt is 538126.0% slower :P
well, at worst
it might be a fluke, maybe we should wait for the next perf run
On average its "only" 348058% slower
how many runs are those averages based on?
@simulacrum would know best. I think it's taking the fastest out of 5 for the smaller benchmarks
Heh at best 2 I think
what? we have a sample of 2?
that does not sound good :)
well, we take the minimum
this is in fact fairly reliable
for most "core" regressions you are actually running something like 2 * 4 (each run type) so you can usually see spurious stuff fairly well
in any case this specific regression has been there for a while I believe
i.e., we have multiple commits backing it up
I've not tried reproducing locally yet though
@simulacrum will perf.rlo test the commits before?
you can see the queue here: https://perf.rust-lang.org/status.html
the length is part of why I suspect this is genuine
I meant: will perf.rlo test the (as of yet untested) commit range, so we can better pinpoint the regression?
the list on the status page probably has the answer :)
Yes, if there's bors commits in between we'll test them eventually
According to the graphs/numbers only baseline incremental and clean is affected that hard.
What's the difference about those and a clean incremental run?
whatever it is is cached by an incremental run
there's descriptions of the variants at the bottom of the compare page
https://github.com/rust-lang/rust/pull/65249 should undo the regression. Thanks @Matthew Jasper and @nnethercote for looking into it!