Wasmtime GitHub notifications bot (Jan 12 2023 at 20:49): Wasmtime GitHub notifications bot (Jan 12 2023 at 20:49):alexcrichton opened PR #5566 from optimize-wiggle to main:
I was working with @abrown today on some performance issues with
wiggleand these are some of the conclusions that we reached. There's a fair amount of low-hanging fruit inwiggleso this was roughly the time that we started picking!
Wasmtime GitHub notifications bot (Jan 12 2023 at 20:49):alexcrichton requested abrown for a review on PR #5566.
Wasmtime GitHub notifications bot (Jan 12 2023 at 20:49):alexcrichton has marked PR #5566 as ready for review.
Wasmtime GitHub notifications bot (Jan 12 2023 at 21:02):abrown submitted PR review.
Wasmtime GitHub notifications bot (Jan 12 2023 at 21:17):alexcrichton updated PR #5566 from optimize-wiggle to main.
Wasmtime GitHub notifications bot (Jan 12 2023 at 21:17):alexcrichton has enabled auto merge for PR #5566.
Wasmtime GitHub notifications bot (Jan 12 2023 at 21:25):bjorn3 submitted PR review.
Wasmtime GitHub notifications bot (Jan 12 2023 at 21:25):bjorn3 created PR review comment:
It isn't AFAICT. You did need an atomic memcpy to prevent a data race, which rust doesn't currently provide. I believe there is an open RFC for it though and you can implement it in assembly.
Wasmtime GitHub notifications bot (Jan 12 2023 at 21:49):alexcrichton merged PR #5566.
Wasmtime GitHub notifications bot (Jan 12 2023 at 22:54):alexcrichton submitted PR review.
Wasmtime GitHub notifications bot (Jan 12 2023 at 22:54):alexcrichton created PR review comment:
Personally I don't think this is a clear yes or no one way or the other. The documentation doesn't explicitly say anything, the documentation of LLVM about their memcpy intrinsic doesn't say anything to this effect, nor does
memcpy-the-function say anything. The linked issue here is the best place to discuss this and its future resolution.
Last updated: Dec 23 2024 at 12:05 UTC