Add an InstSimplify for repetitive array expressions #135274

saethlin · 2025-01-09T02:13:48Z

I noticed in #135068 (comment) that GVN's implementation of this same transform was quite profitable on the deep-vector benchmark. But of course GVN doesn't run in unoptimized builds, so this is my attempt to write a version of this transform that benefits the deep-vector case and is fast enough to run in InstSimplify.

The benchmark suite indicates that this is effective.

saethlin · 2025-01-09T02:14:00Z

@bors try @rust-timer queue

saethlin · 2025-01-09T03:04:47Z

@bors try

Add an InstSimplify for repetitive array expressions r? ghost Let's see if this even perfs well.

bors · 2025-01-09T03:05:59Z

⌛ Trying commit d767dd8 with merge 9811408f4eb6418d113324d8e3609a0a84b54664...

bors · 2025-01-09T04:55:05Z

☀️ Try build successful - checks-actions
Build commit: 9811408 (9811408f4eb6418d113324d8e3609a0a84b54664)

rust-timer · 2025-01-09T06:36:09Z

Finished benchmarking commit (9811408): comparison URL.

Overall result: ✅ improvements - no action needed

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

@bors rollup=never
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

This is the most reliable metric that we have; it was used to determine the overall result at the top of this comment. However, even this metric can sometimes exhibit noise.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-14.6%	[-27.9%, -1.2%]	9
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

Results (primary 2.8%, secondary -11.7%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.8%	[2.8%, 2.8%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-11.7%	[-18.3%, -0.8%]	6
All ❌✅ (primary)	2.8%	[2.8%, 2.8%]	1

Cycles

Results (secondary -28.4%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-28.4%	[-33.2%, -10.9%]	5
All ❌✅ (primary)	-	-	0

Binary size

Results (secondary -25.2%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-25.2%	[-25.2%, -25.2%]	5
All ❌✅ (primary)	-	-	0

Bootstrap: 764.798s -> 763.387s (-0.18%)
Artifact size: 325.77 MiB -> 325.71 MiB (-0.02%)

rustbot · 2025-01-09T22:29:41Z

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

saethlin · 2025-01-09T22:30:43Z

The CI error is tracked here: https://rust-lang.zulipchat.com/#narrow/channel/242791-t-infra/topic/ERROR.3A.20failed.20to.20download.20llvm.20from.20ci

saethlin · 2025-01-09T22:30:49Z

r? mir-opt

compiler-errors · 2025-01-09T23:04:09Z

compiler/rustc_mir_transform/src/instsimplify.rs

+            let Operand::Constant(field) = field else {
+                return false;
+            };
+            let field = field.const_.eval(self.tcx, self.typing_env, field.span);


Would be cool to see more tests exercising this eval call. Specifically, maybe something like [CONST, CONST, ...] for some global const, something with ZSTs, something gauging whether or not it works with things like nested slices/string lits (probably doesn't? const eval doesn't deduplicate allocations like that right?)

The optimization doesn't work on ZSTs. Ultimately that doesn't matter because RemoveZsts will run before InstSimplify and completely eliminate the statement we would optimize.

But it makes me a bit worried that this doesn't work with ZSTs because we generate a local per array element for ZSTs, but we don't for a numeric literal, and this optimization relies on generating an Operand::Const, not a local per array element that we then Move from.

compiler-errors

ty for the tests

r=me

saethlin · 2025-01-10T02:03:11Z

r? compiler-errors
@bors r=compiler-errors

bors · 2025-01-10T02:03:13Z

📌 Commit a285d20 has been approved by compiler-errors

It is now in the queue for this repository.

bors · 2025-01-10T02:03:14Z

🌲 The tree is currently closed for pull requests below priority 1000. This pull request will be tested once the tree is reopened.

bors · 2025-01-11T06:33:59Z

⌛ Testing commit a285d20 with merge 7e4077d...

bors · 2025-01-11T09:15:59Z

☀️ Test successful - checks-actions
Approved by: compiler-errors
Pushing 7e4077d to master...

rust-timer · 2025-01-11T13:08:01Z

Finished benchmarking commit (7e4077d): comparison URL.

Overall result: ✅ improvements - no action needed

@rustbot label: -perf-regression

Instruction count

This is the most reliable metric that we have; it was used to determine the overall result at the top of this comment. However, even this metric can sometimes exhibit noise.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-14.7%	[-27.9%, -1.3%]	9
All ❌✅ (primary)	-	-	0

Max RSS (memory usage)

Results (secondary -12.1%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-12.1%	[-18.1%, -3.4%]	6
All ❌✅ (primary)	-	-	0

Cycles

Results (secondary -18.2%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	3.2%	[3.2%, 3.2%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-21.2%	[-33.7%, -3.2%]	7
All ❌✅ (primary)	-	-	0

Binary size

Results (secondary -25.2%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-25.2%	[-25.2%, -25.1%]	5
All ❌✅ (primary)	-	-	0

Bootstrap: 763.776s -> 765.598s (0.24%)
Artifact size: 326.00 MiB -> 325.92 MiB (-0.02%)

Use Box::new instead of the box_new intrinsic in vec! expansion r? ghost This should perf differently since rust-lang#135274 landed. Some UI tests are failing locally and seem to indicate a diagnostic regression that I'll fix if this perfs well.

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 Jan 9, 2025

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 Jan 9, 2025

bors added a commit to rust-lang-ci/rust that referenced this pull request Jan 9, 2025

Auto merge of rust-lang#135274 - saethlin:array-repeats, r=<try>

9811408

Add an InstSimplify for repetitive array expressions r? ghost Let's see if this even perfs well.

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 Jan 9, 2025

saethlin force-pushed the array-repeats branch from d767dd8 to be5157b Compare January 9, 2025 22:23

This comment has been minimized.

Sign in to view

saethlin added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-experimental Status: Ongoing experiment that does not require reviewing and won't be merged in its current state. labels Jan 9, 2025

saethlin marked this pull request as ready for review January 9, 2025 22:29

rustbot assigned oli-obk Jan 9, 2025

compiler-errors requested changes Jan 9, 2025

View reviewed changes

saethlin force-pushed the array-repeats branch from be5157b to 1c9902e Compare January 10, 2025 00:50

This comment has been minimized.

Sign in to view

Add an InstSimplify for repetitive array expressions

a285d20

saethlin force-pushed the array-repeats branch from 1c9902e to a285d20 Compare January 10, 2025 01:00

compiler-errors approved these changes Jan 10, 2025

View reviewed changes

rustbot assigned compiler-errors Jan 10, 2025

rustbot unassigned oli-obk Jan 10, 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 Jan 10, 2025

bors added the merged-by-bors This PR was explicitly merged by bors. label Jan 11, 2025

bors merged commit 7e4077d into rust-lang:master Jan 11, 2025
7 checks passed

rustbot added this to the 1.86.0 milestone Jan 11, 2025

bors mentioned this pull request Jan 11, 2025

[mir-opt] simplify Repeats that don't actually repeat the operand #135322

Open

saethlin deleted the array-repeats branch January 11, 2025 12:11

saethlin mentioned this pull request Jan 11, 2025

Use Box::new instead of the box_new intrinsic in vec! expansion #135365

Draft

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add an InstSimplify for repetitive array expressions #135274

Add an InstSimplify for repetitive array expressions #135274

saethlin commented Jan 9, 2025 •

edited

Loading

saethlin commented Jan 9, 2025

This comment has been minimized.

saethlin commented Jan 9, 2025

bors commented Jan 9, 2025

bors commented Jan 9, 2025

This comment has been minimized.

rust-timer commented Jan 9, 2025

This comment has been minimized.

rustbot commented Jan 9, 2025

saethlin commented Jan 9, 2025

saethlin commented Jan 9, 2025

compiler-errors Jan 9, 2025

saethlin Jan 10, 2025 •

edited

Loading

This comment has been minimized.

compiler-errors left a comment

saethlin commented Jan 10, 2025

bors commented Jan 10, 2025

bors commented Jan 10, 2025

bors commented Jan 11, 2025

bors commented Jan 11, 2025

rust-timer commented Jan 11, 2025

Add an InstSimplify for repetitive array expressions #135274

Add an InstSimplify for repetitive array expressions #135274

Conversation

saethlin commented Jan 9, 2025 • edited Loading

saethlin commented Jan 9, 2025

This comment has been minimized.

saethlin commented Jan 9, 2025

bors commented Jan 9, 2025

bors commented Jan 9, 2025

This comment has been minimized.

rust-timer commented Jan 9, 2025

Overall result: ✅ improvements - no action needed

This comment has been minimized.

rustbot commented Jan 9, 2025

saethlin commented Jan 9, 2025

saethlin commented Jan 9, 2025

Choose a reason for hiding this comment

saethlin Jan 10, 2025 • edited Loading

Choose a reason for hiding this comment

This comment has been minimized.

compiler-errors left a comment

Choose a reason for hiding this comment

saethlin commented Jan 10, 2025

bors commented Jan 10, 2025

bors commented Jan 10, 2025

bors commented Jan 11, 2025

bors commented Jan 11, 2025

rust-timer commented Jan 11, 2025

Overall result: ✅ improvements - no action needed

saethlin commented Jan 9, 2025 •

edited

Loading

saethlin Jan 10, 2025 •

edited

Loading