2021-08-25 Triage Log

A few regressions but largely an improvement this week, mostly due to the upgrade to LLVM 13.

Triage done by @simulacrum. Revision range: aa8f27bf4d980023a8b245ceb25a490a18041eb2..33fdb797f59421c7bbecaa4588ed5d7a31a9494a

2 Regressions, 1 Improvements, 2 Mixed; 0 of them in rollups

Regressions

Update the backtrace crate in libstd #88151

  • Large regression in instruction counts (up to 2.6% on incr-patched: u8 3072 builds of issue-46449)
  • Standard regression from an increase in std/core size due to added items.

Add fast path for Path::cmp that skips over long shared prefixes #86898

  • Small improvement in instruction counts (up to -0.3% on incr-unchanged builds of helloworld)
  • Moderate regression in instruction counts (up to 1.2% on incr-patched: println builds of ripgrep)
  • Largely in -doc builds, seems to be a small but real regression.

Improvements

Reenable RemoveZsts #88176

  • Large improvement in instruction counts (up to -4.2% on incr-unchanged builds of deeply-nested-async)

Mixed

Revert “Auto merge of #83417 - erikdesjardins:enableremovezsts, r=oli-obk” #88056

  • Small improvement in instruction counts (up to -0.5% on incr-patched: println builds of cargo)
  • Large regression in instruction counts (up to 4.5% on incr-unchanged builds of deeply-nested-async)
  • Regression is canceled out by the later #88176 (see improvements above).

Upgrade to LLVM 13 #87570

  • Very large improvement in instruction counts (up to -10.4% on incr-unchanged builds of helloworld)
  • Very large regression in instruction counts (up to 9.6% on full builds of match-stress-enum)
  • Overall an improvement, regressions are significantly more limited to just a few cases compared to the improvements seen across the board.

Untriaged Pull Requests