jameysharp opened PR #6214 from jameysharp:select-opts
to bytecodealliance:main
:
I've been thinking about the algebraic relationships between
bmask
and other instructions due to recent PRs, and wrote down a few of them here.These patterns don't seem to appear in any of the Sightglass
bz2
,pulldown-cmark
, andspidermonkey
benchmarks, as this PR has no effect on execution-time instructions retired for those benchmarks.
jameysharp requested abrown for a review on PR #6214.
jameysharp requested wasmtime-compiler-reviewers for a review on PR #6214.
fitzgen submitted PR review:
I think we want to be really careful about including
select_spectre_guard
optimizations in here. Might want to pull them out to their own file where we can argue why they are speculation safe.Also these new optimization cases need tests.
Last updated: Dec 23 2024 at 12:05 UTC