Use regular Vec in BitSet.

[cjgillot]

That code is hot enough for the branch in all accesses to SmallVec to appear in profiles.

[cjgillot]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: f78185f (f78185f71f4f0cd33ef5619fdcf876df1237a01f, parent: ed1d94311e7ed53eabb5667ef577802d88d1aaee)

[rust-timer]

Finished benchmarking commit (f78185f): comparison URL.

Overall result: ❌✅ regressions and improvements - please read the text below

Benchmarking this pull request means it may be perf-sensitive – we'll automatically label it not fit for rolling up. You can override this, but we strongly advise not to, due to possible changes in compiler perf.

Next Steps: If you can justify the regressions found in this try perf run, please do so in sufficient writing along with @rustbot label: +perf-regression-triaged. If not, please fix the regressions and do another perf run. If its results are neutral or positive, the label will be automatically removed.

@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)	0.4%	[0.1%, 0.6%]	18
Regressions ❌ (secondary)	0.4%	[0.1%, 0.8%]	23
Improvements ✅ (primary)	-0.4%	[-0.6%, -0.1%]	25
Improvements ✅ (secondary)	-0.7%	[-3.0%, -0.0%]	73
All ❌✅ (primary)	-0.1%	[-0.6%, 0.6%]	43

Max RSS (memory usage)

Results (primary -2.0%, secondary 2.5%)

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)	4.2%	[0.9%, 6.3%]	16
Improvements ✅ (primary)	-2.0%	[-2.0%, -2.0%]	1
Improvements ✅ (secondary)	-2.0%	[-2.9%, -1.2%]	6
All ❌✅ (primary)	-2.0%	[-2.0%, -2.0%]	1

Cycles

Results (secondary -0.5%)

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)	2.6%	[1.7%, 5.2%]	4
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.0%	[-4.2%, -2.1%]	5
All ❌✅ (primary)	-	-	0

Binary size

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

Bootstrap: 473.894s -> 473.617s (-0.06%)
Artifact size: 388.15 MiB -> 390.42 MiB (0.59%)

[rustbot]

r? @jackh726

rustbot has assigned @jackh726.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[saethlin]

Weird. This was a clear improvement when it was added in #109458.

[jackh726]

Yeah, it's definitely curious that perf is also more green than red in this PR, when the other PR was clearly the other (and, it's not even a clear flip-flop of benchmarks).

I'm sort of wondering if different uses of DenseBitSet have different performance tradeoffs? And over time we use DenseBitSet in different ways?

I'm inclined to land this, given the pretty clear perf results. So r=me. But it might be nice to get a better sense of why the perf results are as they are.

[mati865]

Given the artifacts size changes, could this be somehow related to either inlining from LTO or PGO/BOLT?

[cjgillot]

Weird. This was a clear improvement when it was added in #109458.

That PR is over 2 years old, so I'm not really surprised that the usage pattern of BitSets have changed in the mean time. I'm not sure it's worth trying to find which usage change causes it.

@bors r=jackh726

[bors]

📌 Commit 0432322 has been approved by jackh726

It is now in the queue for this repository.