valtree performance tuning #136593

lukas-code · 2025-02-05T17:45:09Z

Summary: This PR makes type checking of code with many type-level constants faster.

After #136180 was merged, we observed a small perf regression (#136318 (comment)). This happened because that PR introduced additional copies in the fast reject code path for consts, which is very hot for certain crates:

rust/compiler/rustc_type_ir/src/fast_reject.rs

Lines 486 to 487 in 6c1d960

    
           ty::ConstKind::Value(lhs_val) => match rhs.kind() { 
        
               ty::ConstKind::Value(rhs_val) => lhs_val.valtree() == rhs_val.valtree(),

This PR improves the performance again by properly interning the valtrees so that copying and comparing them becomes faster. This will become especially useful with feature(adt_const_params), so the fast reject code doesn't have to do a deep compare of the valtrees.

Note that we can't just compare the interned consts themselves in the fast reject, because sometimes 'static lifetimes in the type are be replaced with inference variables (due to canonicalization) on one side but not the other.

A less invasive alternative that I considered is simply avoiding copies introduced by #136180 and comparing the valtrees it in-place (see commit: 9e91e50 / perf results: #136593 (comment)), however that was still measurably slower than interning.

There are some minor regressions in secondary benchmarks: These happen due to changes in memory allocations and seem acceptable to me. The crates that make heavy use of valtrees show no significant changes in memory usage.

lukas-code · 2025-02-05T17:45:33Z

@bors try @rust-timer queue

[perf] try to mitigate regression in fast reject for `ty::Value` rust-lang#136318 (comment) r? ghost

bors · 2025-02-05T17:46:46Z

⌛ Trying commit dc8caa3 with merge 95bc439...

bors · 2025-02-05T19:36:30Z

☀️ Try build successful - checks-actions
Build commit: 95bc439 (95bc439e0cc5e28a3a245d530f8acb3b84d250ac)

rust-timer · 2025-02-06T12:37:01Z

Finished benchmarking commit (95bc439): comparison URL.

Overall result: no relevant changes - 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 benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

Results (primary -2.0%, secondary -1.1%)

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)	6.3%	[6.3%, 6.3%]	1
Improvements ✅ (primary)	-2.0%	[-2.0%, -2.0%]	1
Improvements ✅ (secondary)	-3.5%	[-3.8%, -3.4%]	3
All ❌✅ (primary)	-2.0%	[-2.0%, -2.0%]	1

Cycles

Results (secondary 2.5%)

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)	2.5%	[2.5%, 2.5%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Binary size

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

Bootstrap: 778.541s -> 779.446s (0.12%)
Artifact size: 328.82 MiB -> 328.84 MiB (0.01%)

lukas-code · 2025-02-07T04:56:48Z

@bors try @rust-timer queue

[perf] try to mitigate regression in fast reject for `ty::Value` rust-lang#136318 (comment) r? ghost

bors · 2025-02-07T04:57:59Z

⌛ Trying commit bc51bb2 with merge 9e91e50...

bors · 2025-02-07T06:49:27Z

☀️ Try build successful - checks-actions
Build commit: 9e91e50 (9e91e50ac5920f0b9b4a3b1e0880c85336ba5c64)

rust-timer · 2025-02-07T08:58:19Z

Finished benchmarking commit (9e91e50): 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)	-1.6%	[-2.6%, -0.1%]	14
Improvements ✅ (secondary)	-0.3%	[-0.3%, -0.3%]	1
All ❌✅ (primary)	-1.6%	[-2.6%, -0.1%]	14

Max RSS (memory usage)

Results (primary -0.9%, secondary -2.5%)

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)	1.1%	[1.1%, 1.1%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-3.0%	[-3.0%, -3.0%]	1
Improvements ✅ (secondary)	-2.5%	[-2.5%, -2.5%]	1
All ❌✅ (primary)	-0.9%	[-3.0%, 1.1%]	2

Cycles

Results (primary -2.1%)

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)	-2.1%	[-3.0%, -1.6%]	9
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-2.1%	[-3.0%, -1.6%]	9

Binary size

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

Bootstrap: 780.473s -> 782.144s (0.21%)
Artifact size: 329.02 MiB -> 328.96 MiB (-0.02%)

lukas-code · 2025-02-07T19:18:04Z

@bors try @rust-timer queue

[perf] try to mitigate regression in fast reject for `ty::Value` rust-lang#136318 (comment) r? ghost

bors · 2025-02-07T19:19:16Z

⌛ Trying commit cd733ad with merge b50e910...

rustbot · 2025-02-11T21:05:28Z

Some changes occurred in rustc_ty_utils::consts.rs

cc @BoxyUwU

The rustc-dev-guide subtree was changed. If this PR only touches the dev guide consider submitting a PR directly to rust-lang/rustc-dev-guide otherwise thank you for updating the dev guide with your changes.

cc @BoxyUwU, @jieyouxu, @Kobzol

Some changes occurred in src/tools/clippy

cc @rust-lang/clippy

Some changes occurred to the CTFE machinery

cc @rust-lang/wg-const-eval

lukas-code · 2025-02-11T21:06:20Z

r? @oli-obk

bors · 2025-02-12T10:38:08Z

☔ The latest upstream changes (presumably #136905) made this pull request unmergeable. Please resolve the merge conflicts.

compiler/rustc_middle/src/ty/consts/valtree.rs

rustbot · 2025-02-12T23:42:22Z

HIR ty lowering was modified

cc @fmease

oli-obk · 2025-02-13T08:58:30Z

@bors r+

bors · 2025-02-13T08:58:34Z

📌 Commit b722d5d has been approved by oli-obk

It is now in the queue for this repository.

bors · 2025-02-13T15:27:33Z

⌛ Testing commit b722d5d with merge c241e14...

bors · 2025-02-13T18:19:20Z

☀️ Test successful - checks-actions
Approved by: oli-obk
Pushing c241e14 to master...

rust-timer · 2025-02-13T20:47:20Z

Finished benchmarking commit (c241e14): 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)	-2.6%	[-5.6%, -0.2%]	20
Improvements ✅ (secondary)	-0.3%	[-0.3%, -0.3%]	1
All ❌✅ (primary)	-2.6%	[-5.6%, -0.2%]	20

Max RSS (memory usage)

Results (secondary -4.0%)

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)	-4.0%	[-4.0%, -4.0%]	1
All ❌✅ (primary)	-	-	0

Cycles

Results (primary -3.2%, secondary -2.1%)

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)	-3.2%	[-5.8%, -2.1%]	18
Improvements ✅ (secondary)	-2.1%	[-2.1%, -2.1%]	1
All ❌✅ (primary)	-3.2%	[-5.8%, -2.1%]	18

Binary size

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

Bootstrap: 792.804s -> 788.486s (-0.54%)
Artifact size: 347.85 MiB -> 347.76 MiB (-0.03%)

valtree performance tuning Summary: This PR makes type checking of code with many type-level constants faster. After rust-lang/rust#136180 was merged, we observed a small perf regression (rust-lang/rust#136318 (comment)). This happened because that PR introduced additional copies in the fast reject code path for consts, which is very hot for certain crates: /~https://github.com/rust-lang/rust/blob/6c1d960d88dd3755548b3818630acb63fa98187e/compiler/rustc_type_ir/src/fast_reject.rs#L486-L487 This PR improves the performance again by properly interning the valtrees so that copying and comparing them becomes faster. This will become especially useful with `feature(adt_const_params)`, so the fast reject code doesn't have to do a deep compare of the valtrees. Note that we can't just compare the interned consts themselves in the fast reject, because sometimes `'static` lifetimes in the type are be replaced with inference variables (due to canonicalization) on one side but not the other. A less invasive alternative that I considered is simply avoiding copies introduced by rust-lang/rust#136180 and comparing the valtrees it in-place (see commit: rust-lang/rust@9e91e50 / perf results: rust-lang/rust#136593 (comment)), however that was still measurably slower than interning. There are some minor regressions in secondary benchmarks: These happen due to changes in memory allocations and seem acceptable to me. The crates that make heavy use of valtrees show no significant changes in memory usage.

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 Feb 5, 2025