Small performance wins for Rc & Arc

[tgross35]

This branch is a perf experiment based on #89132 (Allocator API for Rc and Arc) with more inline indicators.

The results were successful so after #89132 lands, I will rebase this and open for review.

[rustbot]

r? @Mark-Simulacrum

(rustbot has picked a reviewer for you, use r? to override)

[Kobzol]

@bors try @rust-timer queue

[bors]

⌛ Trying commit f7fcb93df6891e3b50e82a156a1b033a27b4d60f with merge 75d9a85e6d2bc257dee9d3c14a06743525de77ff...

[bors]

☀️ Try build successful - checks-actions
Build commit: 75d9a85e6d2bc257dee9d3c14a06743525de77ff (75d9a85e6d2bc257dee9d3c14a06743525de77ff)

[bors]

☀️ Try build successful - checks-actions
Build commit: 75d9a85e6d2bc257dee9d3c14a06743525de77ff (75d9a85e6d2bc257dee9d3c14a06743525de77ff)

[rust-timer]

Finished benchmarking commit (75d9a85e6d2bc257dee9d3c14a06743525de77ff): comparison URL.

Overall result: ❌✅ regressions and improvements - 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.

Next Steps: If you can justify the regressions found in this try perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please fix the regressions and do another perf run. If the next run shows neutral or positive results, the label will be automatically removed.

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

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	0.4%	[0.3%, 0.5%]	3
Improvements ✅ (primary)	-0.5%	[-0.7%, -0.4%]	5
Improvements ✅ (secondary)	-0.7%	[-1.9%, -0.3%]	10
All ❌✅ (primary)	-0.5%	[-0.7%, -0.4%]	5

Max RSS (memory usage)

Results

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)	1.4%	[0.5%, 2.2%]	3
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-1.9%	[-4.0%, -0.5%]	5
Improvements ✅ (secondary)	-1.4%	[-1.6%, -1.3%]	2
All ❌✅ (primary)	-0.6%	[-4.0%, 2.2%]	8

Cycles

Results

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)	-0.4%	[-0.4%, -0.4%]	2
Improvements ✅ (secondary)	-2.1%	[-2.1%, -2.1%]	1
All ❌✅ (primary)	-0.4%	[-0.4%, -0.4%]	2

Binary size

Results

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)	0.2%	[0.0%, 0.5%]	44
Regressions ❌ (secondary)	1.5%	[1.5%, 1.5%]	4
Improvements ✅ (primary)	-0.1%	[-0.1%, -0.0%]	4
Improvements ✅ (secondary)	-3.0%	[-3.5%, -1.5%]	4
All ❌✅ (primary)	0.2%	[-0.1%, 0.5%]	48

Bootstrap: 661.126s -> 661.703s (0.09%)

[tgross35]

Okay, I cleaned this up so it should be mergable, so I'll mark it ready for review.

The changes here should be about the same as what produced the perf results above. The performance wins are fairly significant, but the binary size increases are also notable. I'll leave it up to the reviewer what the call is on this.

This may merit a perf rerun since the base changed significantly (the above run included adding the allocator API changes, since then the Arc Allocator API has been merged).

See also #114446 where I'm trying to figure out some inlining things for the layout calculation (but I will keep that PR separate from this one since that one touches more things)

cc @TimDiekmann, this is the followup to adding the allocator API
@rustbot ready

[Mark-Simulacrum]

@bors try @rust-timer queue

[bors]

⌛ Trying commit 864445b with merge 0ef074b4e9cbe42f2c6a4e5f3c529e371fdae41f...

[bors]

☀️ Try build successful - checks-actions
Build commit: 0ef074b4e9cbe42f2c6a4e5f3c529e371fdae41f (0ef074b4e9cbe42f2c6a4e5f3c529e371fdae41f)

[bors]

☀️ Try build successful - checks-actions
Build commit: 0ef074b4e9cbe42f2c6a4e5f3c529e371fdae41f (0ef074b4e9cbe42f2c6a4e5f3c529e371fdae41f)

[tgross35]

It looks like perf maybe didn’t get queued, could you retry @Mark-Simulacrum?

[Kobzol]

@rust-timer build 0ef074b4e9cbe42f2c6a4e5f3c529e371fdae41f

[rust-timer]

Finished benchmarking commit (0ef074b4e9cbe42f2c6a4e5f3c529e371fdae41f): comparison URL.

Overall result: ❌✅ regressions and improvements - 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.

Next Steps: If you can justify the regressions found in this try perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please fix the regressions and do another perf run. If the next run shows neutral or positive results, the label will be automatically removed.

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

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	1.2%	[1.1%, 1.2%]	2
Regressions ❌ (secondary)	1.2%	[0.5%, 1.4%]	5
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.8%	[-2.1%, -0.3%]	6
All ❌✅ (primary)	1.2%	[1.1%, 1.2%]	2

Max RSS (memory usage)

Results

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)	4.5%	[1.3%, 7.0%]	3
Regressions ❌ (secondary)	2.3%	[2.3%, 2.3%]	1
Improvements ✅ (primary)	-7.2%	[-7.2%, -7.2%]	1
Improvements ✅ (secondary)	-2.4%	[-2.9%, -1.9%]	2
All ❌✅ (primary)	1.6%	[-7.2%, 7.0%]	4

Cycles

Results

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)	4.3%	[4.3%, 4.3%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.0%	[-2.2%, -1.7%]	2
All ❌✅ (primary)	-	-	0

Binary size

Results

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)	0.0%	[0.0%, 0.0%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.1%	[-0.6%, -0.0%]	7
Improvements ✅ (secondary)	-2.8%	[-3.3%, -1.4%]	4
All ❌✅ (primary)	-0.1%	[-0.6%, 0.0%]	8

Bootstrap: 651.433s -> 652.417s (0.15%)

[tgross35]

Thanks @Kobzol

Well, this time around the results seem a lot more like noise

[Mark-Simulacrum]

I think this isn't worth it given the largely noise results here. Thanks for trying it!