Start benchmarking #122
Merged
Start benchmarking #122
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
madsmtm
force-pushed
the
criterion-benchmarking
branch
6 times, most recently
from
850e969
to
15682cc
Compare
madsmtm
force-pushed
the
criterion-benchmarking
branch
from
30f4c01
to
22dd211
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Important for #104 and #81.
This took wayyy longer than expected, but hey, I guess I learned a lot about Valgrind and such in the process!
Findings:
criterionis pretty good, though I didn't really get the output stability from it that I would have liked; though that's to be expected when measuring wall clock time. Usingcriterion-cpu-time(others:criterion-cycles-per-byte,criterion-linux-perforcriterion-perf-events) I did manage to get the output a bit more stable, but not enough to really say anything for sure.iaioutput varies a lot on my system, in part I think it's the macOS version of Valgrind/Cachegrind that's the issue. Moving to Valgrind/Callgrind (see Use Callgrind instead of Cachegrind bheisler/iai#26) made the situation much better, and combined with a deterministic allocator the output was actually completely stable!But most importantly:
objc_retainAutoreleasedReturnValue+objc_release("optimized caller") is faster than without doing it (even when still using autorelease pools)!This has significant implications on #86!