perf: skip irrelevant foreign impls when building the specialization graph

[xmakro]

View all comments

Skip foreign non-blanket impls that can't overlap any local impl when building the specialization graph. The call site showed up when profiling and this PR leads to a mean -1.5% instructions perf improvement, see #157281 (comment)

[cjgillot]

@bors try @rust-timer queue

[rust-bors]

💥 Test timed out after 21600s

[xmakro]

@cjgillot could you please retry rust-timer

[Kobzol]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: 55fd877 (55fd87784ebf6c3bcb3d43089182c710005c85d3, parent: 485ec3fbcc12fa14ef6596dabb125ad710499c9e)

[rust-timer]

Finished benchmarking commit (55fd877): comparison URL.

Overall result: ✅ improvements - no action needed

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.

@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)	1.8%	[1.8%, 1.8%]	1
Improvements ✅ (primary)	-1.5%	[-5.8%, -0.2%]	189
Improvements ✅ (secondary)	-3.2%	[-19.8%, -0.1%]	133
All ❌✅ (primary)	-1.5%	[-5.8%, -0.2%]	189

Max RSS (memory usage)

Results (primary -2.2%, secondary -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)	-2.2%	[-5.5%, -0.6%]	112
Improvements ✅ (secondary)	-2.8%	[-5.2%, -0.5%]	69
All ❌✅ (primary)	-2.2%	[-5.5%, -0.6%]	112

Cycles

Results (primary -4.2%, secondary -8.1%)

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)	-	-	0
Improvements ✅ (primary)	-4.2%	[-7.3%, -1.8%]	61
Improvements ✅ (secondary)	-8.1%	[-19.2%, -1.3%]	52
All ❌✅ (primary)	-4.2%	[-7.3%, -1.8%]	61

Binary size

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

Bootstrap: 517.006s -> 513.501s (-0.68%)
Artifact size: 401.43 MiB -> 400.97 MiB (-0.11%)

[xmakro]

@cjgillot would you be comfortable reviewing this PR? or maybe someone from the types team? thanks!

[xmakro]

r? types

[xmakro]

cc @nnethercote who might be interested

[jackh726]

Seems reasonable...

@bors r+

[rust-bors]

📌 Commit 1ddeb9a has been approved by jackh726

It is now in the queue for this repository.

[JonathanBrouwer]

@bors rollup=iffy
Including this in a special perf-sensitive rollup

[JonathanBrouwer]

Verifying that actual perf results after merge match expected results
@rust-timer build 6e2e027

[rust-timer]

Finished benchmarking commit (6e2e027): comparison URL.

Overall result: ✅ improvements - no action needed

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.

@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
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-1.5%	[-6.1%, -0.2%]	209
Improvements ✅ (secondary)	-3.2%	[-19.7%, -0.1%]	137
All ❌✅ (primary)	-1.5%	[-6.1%, -0.2%]	209

Max RSS (memory usage)

Results (primary -2.0%, secondary -2.6%)

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

	mean	range	count
Regressions ❌ (primary)	1.4%	[1.0%, 1.9%]	2
Regressions ❌ (secondary)	2.4%	[0.7%, 4.2%]	2
Improvements ✅ (primary)	-2.0%	[-4.1%, -0.4%]	138
Improvements ✅ (secondary)	-2.7%	[-6.1%, -0.4%]	83
All ❌✅ (primary)	-2.0%	[-4.1%, 1.9%]	140

Cycles

Results (primary -3.5%, secondary -6.4%)

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

	mean	range	count
Regressions ❌ (primary)	0.6%	[0.6%, 0.6%]	1
Regressions ❌ (secondary)	0.8%	[0.6%, 1.1%]	3
Improvements ✅ (primary)	-3.5%	[-7.6%, -0.7%]	85
Improvements ✅ (secondary)	-6.7%	[-20.4%, -0.5%]	70
All ❌✅ (primary)	-3.5%	[-7.6%, 0.6%]	86

Binary size

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

Bootstrap: 489.577s -> 487.42s (-0.44%)
Artifact size: 390.24 MiB -> 390.39 MiB (0.04%)