Automatic #[inline] for trivial wrappers

[saethlin]

r? @ghost

[saethlin]

@bors try @rust-timer queue

[saethlin]

@bors try @rust-timer queue

[saethlin]

@bors try

[bors]

⌛ Trying commit 92837ec with merge c00e9cf...

[bors]

☀️ Try build successful - checks-actions
Build commit: c00e9cf (c00e9cfb5108e1a2c338bd8210525327a2d1277a)

[rust-timer]

Finished benchmarking commit (c00e9cf): 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)	2.6%	[0.2%, 55.1%]	37
Regressions ❌ (secondary)	1.7%	[0.7%, 2.4%]	8
Improvements ✅ (primary)	-0.9%	[-2.3%, -0.3%]	13
Improvements ✅ (secondary)	-0.5%	[-0.6%, -0.5%]	5
All ❌✅ (primary)	1.7%	[-2.3%, 55.1%]	50

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)	3.8%	[0.7%, 6.4%]	6
Regressions ❌ (secondary)	2.7%	[1.7%, 4.2%]	4
Improvements ✅ (primary)	-4.8%	[-9.9%, -1.6%]	6
Improvements ✅ (secondary)	-2.2%	[-3.5%, -0.9%]	3
All ❌✅ (primary)	-0.5%	[-9.9%, 6.4%]	12

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)	4.9%	[0.9%, 56.3%]	19
Regressions ❌ (secondary)	2.1%	[1.7%, 2.5%]	4
Improvements ✅ (primary)	-1.2%	[-2.0%, -0.9%]	6
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	3.4%	[-2.0%, 56.3%]	25

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)	0.5%	[0.0%, 4.2%]	69
Regressions ❌ (secondary)	1.3%	[0.2%, 3.1%]	13
Improvements ✅ (primary)	-0.7%	[-2.3%, -0.0%]	31
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.1%	[-2.3%, 4.2%]	100

Bootstrap: 628.593s -> 628.523s (-0.01%)
Artifact size: 303.98 MiB -> 304.18 MiB (0.07%)

[saethlin]

This may have changed since we now add #[inline] to derived Debug impls
@bors try @rust-timer queue

[bors]

⌛ Trying commit 3c388ba with merge 1dbed32...

[bors]

☀️ Try build successful - checks-actions
Build commit: 1dbed32 (1dbed32e70384502904754e9029f18700d75b030)

[rust-timer]

Finished benchmarking commit (1dbed32): 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)	2.6%	[0.1%, 55.5%]	37
Regressions ❌ (secondary)	1.5%	[0.4%, 2.5%]	9
Improvements ✅ (primary)	-0.8%	[-2.2%, -0.3%]	11
Improvements ✅ (secondary)	-0.7%	[-1.7%, -0.4%]	8
All ❌✅ (primary)	1.8%	[-2.2%, 55.5%]	48

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)	1.6%	[0.4%, 4.7%]	19
Regressions ❌ (secondary)	3.7%	[1.1%, 8.4%]	4
Improvements ✅ (primary)	-2.1%	[-7.2%, -0.5%]	11
Improvements ✅ (secondary)	-2.9%	[-4.9%, -1.3%]	6
All ❌✅ (primary)	0.3%	[-7.2%, 4.7%]	30

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)	3.9%	[0.4%, 56.6%]	23
Regressions ❌ (secondary)	2.2%	[1.1%, 2.8%]	11
Improvements ✅ (primary)	-1.0%	[-2.1%, -0.4%]	9
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	2.5%	[-2.1%, 56.6%]	32

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)	0.5%	[0.0%, 3.1%]	54
Regressions ❌ (secondary)	1.1%	[0.2%, 2.2%]	12
Improvements ✅ (primary)	-0.7%	[-2.5%, -0.0%]	39
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-0.0%	[-2.5%, 3.1%]	93

Bootstrap: 674.395s -> 672.993s (-0.21%)
Artifact size: 311.13 MiB -> 311.35 MiB (0.07%)

[saethlin]

I am leaving the same comment here and on #112255.

I'm closing this because I think this has unsolvable (with our current tooling) compile time performance issues.

I think the performance regressions are driven by increasing the connectivity of the incremental compilation dependency graph. The primary effect of MIR optimizations is to decrease compile times by making less work for LLVM, but if we make builds less incremental in order to produce less IR overall, this is working against itself to improve debug build times.

This tension between MIR optimizations enabled by inlining and incremental build granularity may be solvable by computing the MIR call graph ahead of time and consulting the graph in the MIR inliner to find inlining opportunities that do not significantly increase the connectivity of the graph. But implementing that is beyond me.