Only use dense bitsets in dataflow analyses

[cjgillot]

When a dataflow state has the size close to the number of locals, we should prefer a dense bitset, like we already store locals in a dense vector.
Other occurrences of ChunkedBitSet need to be justified by the size of the dataflow state.

[cjgillot]

@bors try @rust-timer queue

[bors]

⌛ Trying commit f528b4c with merge b5f969f...

[bors]

☀️ Try build successful - checks-actions
Build commit: b5f969f (b5f969f469d2f0079022a76176eed90df14849dd)

[bors]

☀️ Try build successful - checks-actions
Build commit: b5f969f (b5f969f469d2f0079022a76176eed90df14849dd)

[rust-timer]

Finished benchmarking commit (b5f969f): 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)	0.3%	[0.3%, 0.3%]	2
Regressions ❌ (secondary)	0.8%	[0.7%, 0.9%]	4
Improvements ✅ (primary)	-0.9%	[-1.8%, -0.2%]	55
Improvements ✅ (secondary)	-1.1%	[-3.5%, -0.3%]	31
All ❌✅ (primary)	-0.9%	[-1.8%, 0.3%]	57

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)	5.6%	[5.6%, 5.6%]	1
Regressions ❌ (secondary)	5.5%	[1.1%, 11.4%]	11
Improvements ✅ (primary)	-3.0%	[-3.0%, -3.0%]	1
Improvements ✅ (secondary)	-1.3%	[-2.2%, -0.8%]	4
All ❌✅ (primary)	1.3%	[-3.0%, 5.6%]	2

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)	-1.3%	[-1.9%, -0.6%]	17
Improvements ✅ (secondary)	-2.6%	[-3.2%, -2.0%]	7
All ❌✅ (primary)	-1.3%	[-1.9%, -0.6%]	17

Binary size

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

Bootstrap: 632.116s -> 629.721s (-0.38%)
Artifact size: 317.31 MiB -> 316.90 MiB (-0.13%)

[rustbot]

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

[lqd]

cc @nnethercote who introduced the chunked bitsets in #93984

[nnethercote]

#93984 (comment) is the best comment about why I introduced these. In short, it was mostly about reducing peak memory usage, and so all metrics should be considered.

Having said that, interestingly the results here don't look like the inverse of #93984. keccak is no longer in the benchmark suite, and it was the most-affected benchmark, but ones like unicode-normalization and match-stress were also affected strongly.

In #93984 I measured a few crates outside of rustc-perf: http-0.2.6, rust-language-tags-0.3.2, tinyvec-1.5.1, vte-0.10.1. It would be worth measuring them too (or updated versions) along with keccak. If the perf results are good, it would be a nice win for simplicity if ChunkedBitSet can be removed.

[cjgillot]

Looking closer to the massif and dhat profiles, the analyses using a MovePathIndex are responsible for the memory consumption. This is to be expected, as there are many more move paths than locals.

New version, only changing the liveness analyses, not initializedness.
@bors try @rust-timer queue

[bors]

⌛ Trying commit a3e432c with merge dba968f...

[bors]

☀️ Try build successful - checks-actions
Build commit: dba968f (dba968f2264931b18c205da12fb604893b69e439)

[rust-timer]

Finished benchmarking commit (dba968f): 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 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
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.4%	[-0.4%, -0.4%]	1
All ❌✅ (primary)	-	-	0

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)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.1%	[-3.2%, -3.1%]	2
All ❌✅ (primary)	-	-	0

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)	5.9%	[1.8%, 14.5%]	9
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Binary size

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

Bootstrap: 630.239s -> 631.272s (0.16%)
Artifact size: 317.29 MiB -> 317.41 MiB (0.04%)

[cjgillot]

r? @nnethercote

For the cases that use MovePathIndex, a better solution would be to shrink the state space. For borrowck, I don't know if this is possible, but I suspect that dropping Copy or !Drop types for drop elaboration should be possible. In another PR.

[bors]

⌛ Testing commit d67d907 with merge 941b825...

[bors]

💔 Test failed - checks-actions

[cjgillot]

@bors r=nnethercote

[bors]

📌 Commit e07ffe9 has been approved by nnethercote

It is now in the queue for this repository.

[bors]

⌛ Testing commit e07ffe9 with merge 05d9c16...

[bors]

💥 Test timed out

[rust-log-analyzer]

A job failed! Check out the build log: (web) (plain)

Click to see the possible cause of the failure (guessed by this bot)

[cjgillot]

@bors retry

[bors]

⌛ Testing commit e07ffe9 with merge 0e42435...

[bors]

☀️ Test successful - checks-actions
Approved by: nnethercote
Pushing 0e42435 to master...

[rust-timer]

Finished benchmarking commit (0e42435): comparison URL.

Overall result: ✅ improvements - no action needed

@rustbot label: -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
Improvements ✅ (primary)	-0.4%	[-0.5%, -0.2%]	5
Improvements ✅ (secondary)	-0.8%	[-1.1%, -0.4%]	2
All ❌✅ (primary)	-0.4%	[-0.5%, -0.2%]	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)	-	-	0
Regressions ❌ (secondary)	4.8%	[4.8%, 4.8%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.6%	[-2.6%, -2.6%]	1
All ❌✅ (primary)	-	-	0

Cycles

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

Binary size

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

Bootstrap: 663.309s -> 661.891s (-0.21%)
Artifact size: 308.35 MiB -> 308.33 MiB (-0.01%)