rustdoc: instrument call to lib_embargo_visit_item #147452
Conversation
rustbot
added
S-waiting-on-author
|
@bors2 try @rust-timer queue |
This comment has been minimized.
This comment has been minimized.
…, r=<try> rustdoc: instrument call to lib_embargo_visit_item
This comment has been minimized.
This comment has been minimized.
rustbot
added
the
S-waiting-on-perf
This comment has been minimized.
This comment has been minimized.
|
Finished benchmarking commit (3ddbf8e): comparison URL. Overall result: ❌ regressions - no action neededBenchmarking this pull request means it may be perf-sensitive – we'll automatically label it not fit for rolling up. You can override this, but we strongly advise not to, due to possible changes in compiler perf. @bors rollup=never Instruction countOur most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.
Max RSS (memory usage)This benchmark run did not return any relevant results for this metric. CyclesResults (secondary 1.9%)A less reliable metric. May be of interest, but not used to determine the overall result above.
Binary sizeThis benchmark run did not return any relevant results for this metric. Bootstrap: 473.216s -> 472.682s (-0.11%) |
rustbot
removed
the
S-waiting-on-perf
|
for crates with many dependancies (ripgrep, large-workspace), this phase accounts for around 1.5% of runtime. we also don't have any truly massive doc bundles like zed in the test suite. I think a 0.5%-1% improvement is likely possible and may be worth pursuing. this could be done either by via parallelism or by only generating that data on demand. |
|
well, |
I had a thought randomly that we could reduce the amount of times lib_embargo_visit_item is called by only calling it when we want to get the effective visibility of an item in a given crate, since DefId contains a CrateNum. However, this could require some fairly complex locking, so I thought I should instrument this pass to get a ceiling of how much performance gain this could give.
If this would make a difference ever, it would be in crates with a large number of dependencies that are never referenced from the docs (no intra-doc links, trait impls, or re-exports)