2020-10-27 Triage Log

Relatively quiet week for performance.

Triage done by @simulacrum. Revision range: 22e6b9c68941996daa45786b4145e6196e51f0f4..824f900a96d752da2d882863c65f9736e5f2b347

0 Regressions, 2 Improvements, 3 Mixed 1 in rollups.

Regressions

Some mixed results, but no PRs with solely regressions this week.

Improvements

Very large improvement in instruction counts (up to -9.9% on incr-full builds of deeply-nested-async-check)

Large improvement in instruction counts (up to -6.2% on incr-full builds of clap-rs-check)
Interesting case where instruction counts show major improvement across the board, but wall times are either a slight regression or largely unchanged (within noise bound).

Mixed

Moderate improvement in instruction counts (up to -1.5% on full builds of match-stress-enum-check)
Moderate regression in instruction counts (up to 1.2% on full builds of unicode_normalization-check)
Unclear as to causes, left nag; will folow-up in a future week.

Moderate regression in instruction counts (up to 2.9% on full builds of regression-31157-debug)
Moderate improvement in instruction counts (up to -1.1% on incr-patched: b9b3e592dd cherry picked builds of style-servo-debug)
Roughly neutral (or an improvement) on wall times; and a long-desired change. Not proposing a revert at this time.

Large improvement in instruction counts (up to -8.1% on full builds of ctfe-stress-4-check)
Moderate regression in instruction counts (up to 4.4% on full builds of cranelift-codegen-opt)
Regression has unclear cause; left a nag for next week.

Nags requiring follow up