Shrink MIR basic-block statement vectors to fit before caching them by joshtriplett · Pull Request #163915 · rust-lang/rust

joshtriplett · 2026-10-07T02:59:59Z

Every MIR basic block allocates a vector for its statements. These statement vectors grow via the usual heuristics, and never get re-shrunk to the correct size for their contents. This holds on to a fair bit of memory.

Shrink them to fit right before we cache them.

For an aws-sdk-ec2 check, this saves ~80 MiB of peak max-RSS (~1.45%), at the cost of 0.3% in instructions.

r? nnethercote

Every MIR basic block allocates a vector for its statements. These statement vectors grow via the usual heuristics, and never get re-shrunk to the correct size for their contents. This holds on to a fair bit of memory. Shrink them to fit right before we cache them. For an aws-sdk-ec2 check, this saves ~80 MiB of peak max-RSS (~1.45%), at the cost of 0.3% in instructions.

rustbot · 2026-10-07T03:00:03Z

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

joshtriplett · 2026-10-07T03:00:48Z

I tested aws-sdk-ec2 because it's not in the perf suite, and I tested a few other things directly, but let's see the whole suite:

@bors try @rust-timer queue

Shrink MIR basic-block statement vectors to fit before caching them

rust-bors · 2026-10-07T03:59:47Z

☀️ Try build successful (CI)
Build commit: 23ca711 (23ca711ad1b965faf6de390dc814f97c6c84da9e)
Base parent: 8d1a764 (8d1a76430406c877b35d0b627e7f796dcf0dfeca)

rust-timer · 2026-10-07T04:43:12Z

Finished benchmarking commit (23ca711): comparison URL.

Overall result: ❌✅ regressions and improvements - please read:

Benchmarking means the PR may be perf-sensitive. It's automatically marked not fit for rolling up. Overriding is possible but disadvised: it risks changing compiler perf.

Next, please: If you can, justify the regressions found in this try perf run in writing along with @rustbot label: +perf-regression-triaged. If not, fix the regressions and do another perf run. Neutral or positive results will clear the label automatically.

@bors rollup=never rustc-perf
@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)	0.1%	[0.1%, 0.2%]	22
Regressions ❌ (secondary)	0.2%	[0.1%, 0.3%]	22
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.3%	[-0.3%, -0.3%]	1
All ❌✅ (primary)	0.1%	[0.1%, 0.2%]	22

Max RSS (memory usage)

Results (primary -1.3%, secondary -2.0%)

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

	mean	range	count
Regressions ❌ (primary)	2.2%	[2.2%, 2.2%]	1
Regressions ❌ (secondary)	2.9%	[2.1%, 4.0%]	5
Improvements ✅ (primary)	-1.5%	[-2.8%, -0.7%]	25
Improvements ✅ (secondary)	-3.5%	[-7.5%, -0.7%]	16
All ❌✅ (primary)	-1.3%	[-2.8%, 2.2%]	26

Cycles

Results (primary -2.8%, secondary 5.7%)

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)	6.8%	[3.4%, 16.1%]	7
Improvements ✅ (primary)	-2.8%	[-2.8%, -2.8%]	2
Improvements ✅ (secondary)	-2.7%	[-2.7%, -2.7%]	1
All ❌✅ (primary)	-2.8%	[-2.8%, -2.8%]	2

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 489.893s -> 491.828s (0.39%)
Artifact size: 408.60 MiB -> 409.33 MiB (0.18%)

joshtriplett · 2026-10-07T05:29:38Z

The 0.1-0.2% performance regressions are expected, in exchange for the substantial max-RSS savings.

joshtriplett · 2026-10-07T05:29:57Z

@rustbot label: +perf-regression-triaged

panstromek · 2026-10-07T07:46:23Z

The 0.1-0.2% performance regressions are expected, in exchange for the substantial max-RSS savings.

I believe this is a bit of a problem, because people often make the tradeoff in the other direction, so somebody might just do the reverse optimization in the future. MaxRSS is also super noisy so it's a bit difficult to judge that tradeoff very well.

Nevertheless, I agree that we should prioritize memory a bit more. This has come up a few times recently (and on your Path PR as well), so I kicked off a thread on zulip about this: #t-compiler/performance > Prioritizing other metrics (Memory and Disk usage), I want to know what others think.

nnethercote · 2026-10-08T21:10:58Z

@@ -483,6 +483,10 @@ fn mir_promoted(
        Some(MirPhase::Analysis(AnalysisPhase::Initial)),
    );



Can you add a short comment explaining that the small time cost is worth the memory saving? r=me with that

View changes since the review

joshtriplett · 2026-10-09T07:44:43Z

@bors r=nnethercote

rust-bors · 2026-10-09T07:44:46Z

📌 Commit 1b7ba11 has been tentatively approved by nnethercote

It will be put into the queue for this repository once PR CI succeeds.

Shrink MIR basic-block statement vectors to fit before caching them Every MIR basic block allocates a vector for its statements. These statement vectors grow via the usual heuristics, and never get re-shrunk to the correct size for their contents. This holds on to a fair bit of memory. Shrink them to fit right before we cache them. For an aws-sdk-ec2 check, this saves ~80 MiB of peak max-RSS (~1.45%), at the cost of 0.3% in instructions. r? nnethercote

JonathanBrouwer · 2026-10-09T10:18:08Z

@bors rollup=never rustc-perf
Bors didn't catch that command for some reason?

rust-log-analyzer · 2026-10-09T11:58:37Z

The job test-aarch64-msvc-1 failed! Check out the build log: (web) (plain enhanced) (plain)

Click to see the possible cause of the failure (guessed by this bot)

---- [run-make] tests\run-make\clear-error-blank-output stdout ----

error: rmake recipe failed to complete
status: exit code: 101
command: "C:\\a\\rust\\rust\\build\\aarch64-pc-windows-msvc\\test\\run-make\\clear-error-blank-output\\rmake.exe"
stdout: none
--- stderr -------------------------------

thread 'main' (7208) panicked at C:\a\rust\rust\tests\run-make\clear-error-blank-output\rmake.rs:11:64:
called `Result::unwrap()` on an `Err` value: Os { code: 232, kind: BrokenPipe, message: "The pipe is being closed." }
---
Currently active steps:
test::RunMake { test_compiler: Compiler { stage: 2, host: aarch64-pc-windows-msvc, forced_compiler: false }, target: aarch64-pc-windows-msvc } at src\bootstrap\src\core\build_steps\test.rs:2108
test::Compiletest { test_compiler: Compiler { stage: 2, host: aarch64-pc-windows-msvc, forced_compiler: false }, target: aarch64-pc-windows-msvc, mode: run-make, suite: "run-make", path: "tests/run-make", compare_mode: None } at src\bootstrap\src\core\build_steps\test.rs:2108
Build completed unsuccessfully in 1:51:53
make: *** [Makefile:115: ci-msvc-py] Error 1
  local time: Fri Oct  9 04:58:13 PDT 2026
  network time: Fri, 09 Oct 2026 11:58:13 GMT
##[error]Process completed with exit code 2.
##[group]Run echo "disk usage:"
echo "disk usage:"

rust-bors · 2026-10-09T11:59:07Z

💔 Test for 9dbebb6 failed: CI. Failed job:

auto - test-aarch64-msvc-1 (web logs, enhanced plaintext logs)

JonathanBrouwer · 2026-10-09T12:04:00Z

@bors retry

rust-bors · 2026-10-09T16:24:26Z

☀️ Test successful - CI
Approved by: nnethercote
Duration: 3h 6m 49s
Pushing 69bccf0 to main...

github-actions · 2026-10-09T16:27:31Z

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 76c9095 (parent) -> 69bccf0 (this PR)

Test differences

Show 3 test diffs

3 doctest diffs were found. These are ignored, as they are noisy.

Test dashboard

Run

cargo run --manifest-path src/ci/citool/Cargo.toml -- \
    test-dashboard 69bccf03c4733f57482f9691456efcd80f86a5f6 --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

test-x86_64-msvc-1: 1h 43m -> 2h 49m (+64.9%)
dist-i686-linux: 1h 10m -> 1h 44m (+48.0%)
test-x86_64-msvc-ext2: 1h 16m -> 1h 52m (+46.7%)
dist-x86_64-netbsd: 1h 2m -> 1h 31m (+45.9%)
test-x86_64-gnu-aux: 2h 42m -> 1h 29m (-44.9%)
test-x86_64-gnu-stdlib-semver-check: 18m 16s -> 10m 31s (-42.4%)
dist-android: 22m 19s -> 31m 47s (+42.4%)
test-x86_64-gnu-stable: 1h 33m -> 2h 13m (+42.4%)
test-aarch64-gnu: 2h 9m -> 1h 15m (-41.6%)
test-pr-check-2: 40m 27s -> 23m 49s (-41.1%)

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 · 2026-10-09T17:23:38Z

Finished benchmarking commit (69bccf0): comparison URL.

Overall result: ❌✅ regressions and improvements - please read:

Our benchmarks found a performance regression caused by this PR.
This might be an actual regression, but it can also be just noise.

Next Steps:

If the regression was expected or you think it can be justified,
please write a comment with sufficient written justification, and add
@rustbot label: +perf-regression-triaged to it, to mark the regression as triaged.
If you think that you know of a way to resolve the regression, try to create
a new PR with a fix for the regression.
If you do not understand the regression or you think that it is just noise,
you can ask the @rust-lang/wg-compiler-performance working group for help (members of this group
were already notified of this PR).

@rustbot label: +perf-regression
cc @rust-lang/wg-compiler-performance

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)	0.1%	[0.1%, 0.2%]	6
Regressions ❌ (secondary)	0.1%	[0.1%, 0.3%]	10
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.1%	[-0.1%, -0.1%]	3
All ❌✅ (primary)	0.1%	[0.1%, 0.2%]	6

Max RSS (memory usage)

Results (primary -1.3%, secondary -3.4%)

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

	mean	range	count
Regressions ❌ (primary)	1.9%	[1.9%, 1.9%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-1.5%	[-2.2%, -0.9%]	12
Improvements ✅ (secondary)	-3.4%	[-6.7%, -0.8%]	15
All ❌✅ (primary)	-1.3%	[-2.2%, 1.9%]	13

Cycles

Results (primary -2.8%)

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)	-2.8%	[-2.8%, -2.8%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-2.8%	[-2.8%, -2.8%]	1

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 490.84s -> 487.936s (-0.59%)
Artifact size: 406.48 MiB -> 406.49 MiB (0.00%)

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Oct 7, 2026

rustbot assigned nnethercote Oct 7, 2026

This comment has been minimized.

Sign in to view

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Oct 7, 2026

This comment has been minimized.

Sign in to view

rust-bors Bot pushed a commit that referenced this pull request Oct 7, 2026

Auto merge of #163915 - joshtriplett:mir-shrink-to-fit, r=<try>

23ca711

Shrink MIR basic-block statement vectors to fit before caching them

This comment has been minimized.

Sign in to view

rustbot added perf-regression Performance regression. and removed S-waiting-on-perf Status: Waiting on a perf run to be completed. labels Oct 7, 2026

rustbot added the perf-regression-triaged The performance regression has been triaged. label Oct 7, 2026

nnethercote reviewed Oct 8, 2026

View reviewed changes

nnethercote added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Oct 8, 2026

Add a comment on the MIR shrink_to_fit of basic blocks

1b7ba11

rust-bors Bot 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-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Oct 9, 2026

This comment has been minimized.

Sign in to view

rust-bors Bot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. labels Oct 9, 2026

rust-bors Bot 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 Oct 9, 2026

This comment has been minimized.

Sign in to view

rust-bors Bot added merged-by-bors This PR was explicitly merged by bors. and removed S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. labels Oct 9, 2026

rust-bors Bot merged commit 69bccf0 into rust-lang:main Oct 9, 2026
15 checks passed

rustbot added this to the 1.101.0 milestone Oct 9, 2026

joshtriplett deleted the mir-shrink-to-fit branch October 9, 2026 17:41

		@@ -483,6 +483,10 @@ fn mir_promoted(
		Some(MirPhase::Analysis(AnalysisPhase::Initial)),
		);

Conversation

joshtriplett commented Oct 7, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Oct 7, 2026

Uh oh!

joshtriplett commented Oct 7, 2026

Uh oh!

This comment has been minimized.

This comment has been minimized.

rust-bors Bot commented Oct 7, 2026

Uh oh!

This comment has been minimized.

rust-timer commented Oct 7, 2026

Overall result: ❌✅ regressions and improvements - please read:

Uh oh!

joshtriplett commented Oct 7, 2026

Uh oh!

joshtriplett commented Oct 7, 2026

Uh oh!

panstromek commented Oct 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

nnethercote Oct 8, 2026 • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

joshtriplett commented Oct 9, 2026

Uh oh!

rust-bors Bot commented Oct 9, 2026

Uh oh!

This comment has been minimized.

JonathanBrouwer commented Oct 9, 2026

Uh oh!

rust-log-analyzer commented Oct 9, 2026

Uh oh!

rust-bors Bot commented Oct 9, 2026

Uh oh!

JonathanBrouwer commented Oct 9, 2026

Uh oh!

This comment has been minimized.

rust-bors Bot commented Oct 9, 2026

Uh oh!

Uh oh!

github-actions Bot commented Oct 9, 2026

Test differences

Job duration changes

Uh oh!

rust-timer commented Oct 9, 2026

Overall result: ❌✅ regressions and improvements - please read:

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

joshtriplett commented Oct 7, 2026 •

edited by rustbot

Loading

panstromek commented Oct 7, 2026 •

edited

Loading

nnethercote Oct 8, 2026 •

edited by rustbot

Loading