Stream: t-compiler/wg-self-profile

Topic: overhead


nikomatsakis (May 07 2019 at 20:17, on Zulip):

Also, @mw and I were talking earlier and discussing the "perf integration". Do we have any idea of "instruction count ovehead" that enabling this feature brings? Memory overhead, I understand, is substantial.

Wesley Wiser (May 07 2019 at 20:18, on Zulip):

mw had a PR where we measured it

Wesley Wiser (May 07 2019 at 20:18, on Zulip):

/me looking

Wesley Wiser (May 07 2019 at 20:20, on Zulip):

https://perf.rust-lang.org/compare.html?start=e4e032a0ae82d7db4f99872ff98626af2941c4a5&end=5539376de500270af54c7741ff8075316d950caf

nikomatsakis (May 07 2019 at 20:21, on Zulip):

Ah, hmm, not that bad.

nikomatsakis (May 07 2019 at 20:32, on Zulip):

My opinion here is that we should probably do it :) at least, it seems so super useful to be able to get this data readily

nikomatsakis (May 07 2019 at 20:33, on Zulip):

but I guess it depends a bit on how consistent the overheads are etc

nikomatsakis (May 07 2019 at 20:33, on Zulip):

(I'm doing some local perf runs and already finding the results very interesting)

nikomatsakis (May 07 2019 at 20:36, on Zulip):

also, reasonably consistent run to run

nikomatsakis (May 07 2019 at 20:37, on Zulip):

some small-ish variation

Wesley Wiser (May 07 2019 at 20:38, on Zulip):

That's good to hear! I don't think we currently have any data about how much variability it introduces.

Last update: Nov 15 2019 at 20:25UTC