[perf] `GenericArgs`-related: Change asserts to debug asserts & use more slice interning over iterable interning #142289

fmease · 2025-06-10T09:56:15Z

The 1st commit yields the following perf gains: #142289 (comment).
The 2nd commit might also have a minor positive perf impact, however that one wasn't tested in isolation.

For reference, the initial approach c7e6acc (results: #142289 (comment)) had a lot more changes (apart from what's now contained in commit 1 and 2) which seemed to be perf irrelevant (cf. the partial countercheck in 6f82bf1 (results: #142289 (comment)).

fmease · 2025-06-10T10:32:04Z

@bors2 try @rust-timer queue

rust-bors · 2025-06-10T10:32:07Z

⌛ Trying commit 5ed5135 with merge c7e6acc…

To cancel the try build, run the command @bors2 try cancel.

[perf] Tweak `GenericArgs`-related functions

compiler/rustc_type_ir/src/ty_kind.rs

rust-bors · 2025-06-10T12:57:58Z

☀️ Try build successful (CI)
Build commit: c7e6acc (c7e6accd79d91fe5dec01a81499a08f9db280440)

rust-timer · 2025-06-11T02:46:25Z

Finished benchmarking commit (c7e6acc): comparison URL.

Overall result: ✅ improvements - no action needed

Benchmarking 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
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.1%	[0.1%, 0.1%]	1
Improvements ✅ (primary)	-0.2%	[-0.4%, -0.1%]	14
Improvements ✅ (secondary)	-1.7%	[-5.6%, -0.2%]	11
All ❌✅ (primary)	-0.2%	[-0.4%, -0.1%]	14

Max RSS (memory usage)

Results (primary -9.0%, secondary 2.7%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	2.7%	[2.7%, 2.7%]	1
Improvements ✅ (primary)	-9.0%	[-9.0%, -9.0%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-9.0%	[-9.0%, -9.0%]	1

Cycles

Results (secondary -1.1%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	2.3%	[2.3%, 2.3%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.8%	[-3.0%, -2.6%]	2
All ❌✅ (primary)	-	-	0

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 753.145s -> 754.777s (0.22%)
Artifact size: 372.18 MiB -> 372.15 MiB (-0.01%)

fmease · 2025-06-11T12:00:57Z

I want to check what happens if I only do the assert_matches → debug_assert_matches change since I can't quite tell from the self-profile and the query trace if that's the actual and only reason for the perf improvement we're seeing.

@bors2 try @rust-timer queue

rust-bors · 2025-06-11T12:01:01Z

⌛ Trying commit 642bb6d with merge ce5fbc7…

To cancel the try build, run the command @bors2 try cancel.

[perf] Tweak `GenericArgs`-related functions Result: #142289 (comment)

fmease · 2025-06-11T12:18:20Z

Wait, I want to set parent=last

@bors2 try cancel

rust-bors · 2025-06-11T12:18:23Z

Try build cancelled. Cancelled workflows:

https://github.com/rust-lang/rust/actions/runs/15584270986

fmease · 2025-06-11T12:26:53Z

@bors2 try parent=40daf23eeb711dadf140b2536e67e3ff4c999196 @rust-timer queue

rust-bors · 2025-06-11T12:26:56Z

⌛ Trying commit 642bb6d with merge 7c7eef3…

To cancel the try build, run the command @bors2 try cancel.

[perf] Tweak `GenericArgs`-related functions Result: #142289 (comment)

rust-bors · 2025-06-11T14:55:47Z

☀️ Try build successful (CI)
Build commit: 7c7eef3 (7c7eef3fba42244270022433fc859ce602611e59, parent: 40daf23eeb711dadf140b2536e67e3ff4c999196)

fmease · 2025-06-12T23:33:06Z

compiler/rustc_hir_analysis/src/delegation.rs

-            let method_args =
-                tcx.mk_args_from_iter(trait_args.iter().skip(callee_generics.parent_count));


Just a drive-by change, definitely not relevant or hot.

fmease · 2025-06-12T23:34:11Z

compiler/rustc_middle/src/ty/context.rs

+        debug_assert_matches!(self.def_kind(def_id), DefKind::AssocTy | DefKind::AssocConst);
        let trait_def_id = self.parent(def_id);
-        assert_matches!(self.def_kind(trait_def_id), DefKind::Trait);
+        debug_assert_matches!(self.def_kind(trait_def_id), DefKind::Trait);


The main and most relevant positive perf impact comes from changing these asserts to debug ones. Results: #142289 (comment).

fmease · 2025-06-12T23:42:13Z

compiler/rustc_middle/src/ty/generic_args.rs

@@ -588,7 +588,7 @@ impl<'tcx> GenericArgs<'tcx> {
    }

    pub fn truncate_to(&self, tcx: TyCtxt<'tcx>, generics: &ty::Generics) -> GenericArgsRef<'tcx> {
-        tcx.mk_args_from_iter(self.iter().take(generics.count()))
+        tcx.mk_args(&self[..generics.count()])


This one wasn't tested in isolation (three perf runs are already a lot (I had to narrow down what is and what isn't perf relevant)) but I presume it to be helping since slicing & slice interning should generally be faster than iterating & "iterable interning" and this function is semi hot / lukewarm I'm pretty sure.

The first approach (which contains this change) yielded more improvements (results: #142289 (comment) (e.g., for image)) than the 1st commit (the assert changes) (results: #142289 (comment)), so this one might be the cause?

I could rerun perf for this change / commit only (which seems a bit overkill) or just drop it entirely? WDYT?

fmease · 2025-06-12T23:50:34Z

r? compiler-errors (since you've reacted in this thread) or lcnr or compiler

compiler-errors

cool. i dont think we'll gain much insight by doing individual perf tests of all the tweaks here; the fact that we gain a bit of perf back anyways is good and the changes at least make the code more readable.

compiler-errors · 2025-06-12T23:55:00Z

r=me when ci is green or sooner i dont care much

fmease · 2025-06-13T00:06:23Z

@bors r=compiler-errors

bors · 2025-06-13T00:06:26Z

📌 Commit 3a31f62 has been approved by compiler-errors

It is now in the queue for this repository.

jieyouxu · 2025-06-14T15:57:26Z

@bors p=5 (to get some rollup=nevers through before new rollups)

bors · 2025-06-14T19:42:07Z

⌛ Testing commit 3a31f62 with merge 49a8ba0...

bors · 2025-06-14T22:40:43Z

☀️ Test successful - checks-actions
Approved by: compiler-errors
Pushing 49a8ba0 to master...

github-actions · 2025-06-14T22:43:53Z

What is this?

This is an experimental post-merge analysis report that shows differences in test outcomes between the merged PR and its parent PR.

Comparing cc87afd (parent) -> 49a8ba0 (this PR)

Test differences

No test diffs found

Test dashboard

Run

cargo run --manifest-path src/ci/citool/Cargo.toml -- \
    test-dashboard 49a8ba06848fa8f282fe9055b4178350970bb0ce --output-dir test-dashboard

And then open test-dashboard/index.html in your browser to see an overview of all executed tests.

Job duration changes

mingw-check-1: 1563.8s -> 2005.6s (28.3%)
x86_64-apple-2: 4102.8s -> 5128.8s (25.0%)
x86_64-apple-1: 5863.1s -> 7297.8s (24.5%)
dist-apple-various: 7288.8s -> 6075.0s (-16.7%)
x86_64-rust-for-linux: 2600.0s -> 2982.7s (14.7%)
i686-gnu-2: 5301.8s -> 5941.4s (12.1%)
i686-gnu-nopt-1: 7165.3s -> 7874.7s (9.9%)
aarch64-gnu-debug: 3570.1s -> 3905.2s (9.4%)
aarch64-apple: 4287.1s -> 4684.6s (9.3%)
x86_64-gnu-aux: 5839.9s -> 6368.6s (9.1%)

How to interpret the job duration changes?

Job durations can vary a lot, based on the actual runner instance
that executed the job, system noise, invalidated caches, etc. The table above is provided
mostly for t-infra members, for simpler debugging of potential CI slow-downs.

rust-timer · 2025-06-15T00:07:33Z

Finished benchmarking commit (49a8ba0): comparison URL.

Overall result: ✅ improvements - no action needed

@rustbot label: -perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.2%	[-0.3%, -0.1%]	15
Improvements ✅ (secondary)	-1.2%	[-4.7%, -0.1%]	13
All ❌✅ (primary)	-0.2%	[-0.3%, -0.1%]	15

Max RSS (memory usage)

Results (secondary -2.2%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.2%	[-2.2%, -2.2%]	1
All ❌✅ (primary)	-	-	0

Cycles

This benchmark run did not return any relevant results for this metric.

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 756.218s -> 758.463s (0.30%)
Artifact size: 372.24 MiB -> 372.25 MiB (0.00%)

[perf] `GenericArgs`-related: Change asserts to debug asserts & use more slice interning over iterable interning 1. The 1st commit yields the following perf gains: [#142289 (comment)](rust-lang/rust#142289 (comment)). 2. The 2nd commit might also have a minor positive perf impact, however that one wasn't tested in isolation. For reference, the initial approach rust-lang/rust@c7e6acc (results: rust-lang/rust#142289 (comment)) had a lot more changes (apart from what's now contained in commit 1 and 2) which seemed to be perf irrelevant (cf. the partial countercheck in rust-lang/rust@6f82bf1 (results: rust-lang/rust#142289 (comment)).

rustbot added the T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. label Jun 10, 2025

fmease added the S-experimental Status: Ongoing experiment that does not require reviewing and won't be merged in its current state. label Jun 10, 2025

fmease force-pushed the maybe-perf-gen-args branch from c9e7983 to e2e88ca Compare June 10, 2025 10:10

fmease added the rla-silenced Silences rust-log-analyzer postings to the PR it's added on. label Jun 10, 2025

fmease force-pushed the maybe-perf-gen-args branch from e2e88ca to 5ed5135 Compare June 10, 2025 10:31

fmease changed the title ~~[perf] Tweak GenericArgs::truncate_to and AliasTerm::own_args~~ [perf] Tweak GenericArgs-related functions Jun 10, 2025

This comment has been minimized.

Sign in to view

rust-bors bot added a commit that referenced this pull request Jun 10, 2025

Auto merge of #142289 - fmease:maybe-perf-gen-args, r=<try>

c7e6acc

[perf] Tweak `GenericArgs`-related functions

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Jun 10, 2025

fmease commented Jun 10, 2025

View reviewed changes

compiler/rustc_type_ir/src/ty_kind.rs Outdated Show resolved Hide resolved

This comment has been minimized.

Sign in to view

rustbot removed the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Jun 11, 2025

fmease force-pushed the maybe-perf-gen-args branch 2 times, most recently from 196f52a to 642bb6d Compare June 11, 2025 11:58

This comment has been minimized.

Sign in to view

rust-bors bot added a commit that referenced this pull request Jun 11, 2025

Auto merge of #142289 - fmease:maybe-perf-gen-args, r=<try>

ce5fbc7

[perf] Tweak `GenericArgs`-related functions Result: #142289 (comment)

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Jun 11, 2025

This comment has been minimized.

Sign in to view

rust-bors bot added a commit that referenced this pull request Jun 11, 2025

Auto merge of #142289 - fmease:maybe-perf-gen-args, r=<try>

7c7eef3

[perf] Tweak `GenericArgs`-related functions Result: #142289 (comment)

fmease added 2 commits June 13, 2025 01:16

[perf] Change asserts to debug ones in trait_ref_and_own_args

a58fdd1

Use more slicing and slice interning over iterable interning

3a31f62

fmease force-pushed the maybe-perf-gen-args branch from e5e876a to 3a31f62 Compare June 12, 2025 23:16

fmease removed S-experimental Status: Ongoing experiment that does not require reviewing and won't be merged in its current state. rla-silenced Silences rust-log-analyzer postings to the PR it's added on. labels Jun 12, 2025

fmease changed the title ~~[perf] Tweak GenericArgs-related functions~~ [perf] trait_ref_own_args: Change assert to debug assert and use more slice interning over iterable interning Jun 12, 2025

fmease commented Jun 12, 2025

View reviewed changes

rustbot assigned compiler-errors Jun 12, 2025

fmease marked this pull request as ready for review June 12, 2025 23:50

rustbot added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jun 12, 2025

compiler-errors approved these changes Jun 12, 2025

View reviewed changes

fmease changed the title ~~[perf] trait_ref_own_args: Change assert to debug assert and use more slice interning over iterable interning~~ [perf] trait_ref_own_args: Change asserts to debug asserts and use more slice interning over iterable interning Jun 12, 2025

fmease changed the title ~~[perf] trait_ref_own_args: Change asserts to debug asserts and use more slice interning over iterable interning~~ [perf] GenericArgs-related: Change asserts to debug asserts & use more slice interning over iterable interning Jun 13, 2025

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Jun 13, 2025

bors added the merged-by-bors This PR was explicitly merged by bors. label Jun 14, 2025

bors merged commit 49a8ba0 into rust-lang:master Jun 14, 2025
11 checks passed

rustbot added this to the 1.89.0 milestone Jun 14, 2025

fmease deleted the maybe-perf-gen-args branch June 14, 2025 22:52

		let method_args =
		tcx.mk_args_from_iter(trait_args.iter().skip(callee_generics.parent_count));

[perf] GenericArgs-related: Change asserts to debug asserts & use more slice interning over iterable interning #142289

[perf] GenericArgs-related: Change asserts to debug asserts & use more slice interning over iterable interning #142289

Uh oh!

Conversation

fmease commented Jun 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

fmease commented Jun 10, 2025

Uh oh!

This comment has been minimized.

rust-bors bot commented Jun 10, 2025

Uh oh!

Uh oh!

rust-bors bot commented Jun 10, 2025

Uh oh!

This comment has been minimized.

rust-timer commented Jun 11, 2025

Overall result: ✅ improvements - no action needed

Uh oh!

fmease commented Jun 11, 2025

Uh oh!

This comment has been minimized.

rust-bors bot commented Jun 11, 2025

Uh oh!

fmease commented Jun 11, 2025

Uh oh!

rust-bors bot commented Jun 11, 2025

Uh oh!

fmease commented Jun 11, 2025

Uh oh!

This comment has been minimized.

rust-bors bot commented Jun 11, 2025

Uh oh!

rust-bors bot commented Jun 11, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

fmease Jun 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

fmease commented Jun 12, 2025

Uh oh!

compiler-errors left a comment

Choose a reason for hiding this comment

Uh oh!

compiler-errors commented Jun 12, 2025

Uh oh!

fmease commented Jun 13, 2025

Uh oh!

bors commented Jun 13, 2025

Uh oh!

jieyouxu commented Jun 14, 2025

Uh oh!

bors commented Jun 14, 2025

Uh oh!

bors commented Jun 14, 2025

Uh oh!

Uh oh!

github-actions bot commented Jun 14, 2025

Test differences

Job duration changes

Uh oh!

rust-timer commented Jun 15, 2025

Overall result: ✅ improvements - no action needed

Uh oh!

Uh oh!

[perf] `GenericArgs`-related: Change asserts to debug asserts & use more slice interning over iterable interning #142289

[perf] `GenericArgs`-related: Change asserts to debug asserts & use more slice interning over iterable interning #142289

fmease commented Jun 10, 2025 •

edited

Loading

fmease Jun 12, 2025 •

edited

Loading