[perf] Loan liveness frontier slowness evaluation #112432

lqd · 2023-06-08T20:55:35Z

It doesn't feel that much slower on UI tests, but will probably scale badly on realistic code, so let's see how terribly it performs.

r? @ghost

lqd · 2023-06-08T20:57:42Z

@bors try @rust-timer queue

bors · 2023-06-08T20:57:53Z

⌛ Trying commit fff2a32f5c6e40599f9fcc713323037667f9d7e8 with merge 272c25090e41e0e5d04bfe45e3c04016d4c6b0a6...

bors · 2023-06-08T22:34:38Z

☀️ Try build successful - checks-actions
Build commit: 272c25090e41e0e5d04bfe45e3c04016d4c6b0a6 (272c25090e41e0e5d04bfe45e3c04016d4c6b0a6)

rust-timer · 2023-06-08T23:52:17Z

Finished benchmarking commit (272c25090e41e0e5d04bfe45e3c04016d4c6b0a6): comparison URL.

Overall result: ❌ regressions - 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)	23.3%	[0.2%, 1404.2%]	181
Regressions ❌ (secondary)	48.2%	[0.1%, 1105.8%]	96
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.5%	[-0.6%, -0.4%]	9
All ❌✅ (primary)	23.3%	[0.2%, 1404.2%]	181

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)	-3.4%	[-4.0%, -2.8%]	3
Improvements ✅ (secondary)	-2.0%	[-2.7%, -0.8%]	8
All ❌✅ (primary)	-3.4%	[-4.0%, -2.8%]	3

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)	31.2%	[1.2%, 478.1%]	45
Regressions ❌ (secondary)	156.4%	[2.3%, 502.4%]	13
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	31.2%	[1.2%, 478.1%]	45

Binary size

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

Bootstrap: 648.127s -> 661.135s (2.01%)

lqd · 2023-06-08T23:53:50Z

That's more like it.

lqd · 2023-06-09T18:25:49Z

@bors try @rust-timer queue

bors · 2023-06-09T18:26:00Z

⌛ Trying commit db6a863c8f0a04969e7e628ce2a6ffffde30769f with merge 48ab5baa844649d42689b368717f6140000cd402...

bors · 2023-06-09T20:02:01Z

☀️ Try build successful - checks-actions
Build commit: 48ab5baa844649d42689b368717f6140000cd402 (48ab5baa844649d42689b368717f6140000cd402)

rust-timer · 2023-06-09T21:38:38Z

Finished benchmarking commit (48ab5baa844649d42689b368717f6140000cd402): 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)	1.6%	[0.2%, 9.5%]	24
Regressions ❌ (secondary)	2.8%	[1.8%, 4.0%]	6
Improvements ✅ (primary)	-0.6%	[-0.7%, -0.6%]	3
Improvements ✅ (secondary)	-0.4%	[-0.6%, -0.3%]	8
All ❌✅ (primary)	1.4%	[-0.7%, 9.5%]	27

Max RSS (memory usage)

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

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)	3.2%	[1.1%, 5.6%]	11
Regressions ❌ (secondary)	2.6%	[2.6%, 2.6%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.0%	[-2.0%, -2.0%]	1
All ❌✅ (primary)	3.2%	[1.1%, 5.6%]	11

Binary size

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

Bootstrap: 649.935s -> 649.406s (-0.08%)

lqd · 2023-06-24T00:00:24Z

@bors try @rust-timer queue

Compute NLL loan scopes using the polonius model Draft PR to guide discussion for this monday's types team meeting, where the plan is to read this and review as a group. For a *location-insensitive* analysis (that is, without expressiveness improvements for users yet), this PR implements loans going out of scope using reachability and liveness, rather than checking if the issuing region's values contain a given CFG point. This is equivalent to NLL scopes and computes the same data. There are a few fixmes left, and I'll still work on them before the meeting. A couple of notes: - I started prototyping this a while ago so it follows the structure of the _previous_ NLL `OutOfScopePrecomputer`, before the couple recent refactors. They probably could/should look the same instead. - there are some assumptions about placeholders, free regions, and member constraints that I think hold, and they're documented in the code - this passes all the UI tests with `-Zpolonius=next` -- though the additional work done can be felt on the huge numeric/float tests, but the perf is [not terrible](rust-lang#112432 (comment)) and there are a bunch of ways to improve it in the future. (cc `@matthewjasper` since you asked to be pinged about it) r? `@ghost`

this allows to opt into using the legacy version or the in-tree prototype

lqd · 2023-10-05T16:38:59Z

@bors try @rust-timer queue

bors · 2023-10-05T16:40:09Z

⌛ Trying commit 99e0fc2 with merge 5281eea...

[perf] Loan liveness frontier slowness evaluation It doesn't feel that much slower on UI tests, but will probably scale badly on realistic code, so let's see how terribly it performs. r? `@ghost`

lqd · 2023-10-05T17:46:52Z

the previous run is going to be busted but the force push didn't kill it.

tests are now fixed

@bors try @rust-timer queue

bors · 2023-10-05T17:48:01Z

⌛ Trying commit 8d5cbc6 with merge a2f1897...

[perf] Loan liveness frontier slowness evaluation It doesn't feel that much slower on UI tests, but will probably scale badly on realistic code, so let's see how terribly it performs. r? `@ghost`

bors · 2023-10-05T19:01:56Z

☀️ Try build successful - checks-actions
Build commit: a2f1897 (a2f18979df8c2af00fed70d11ae69af0d324ae6d)

rust-timer · 2023-10-05T21:37:47Z

Finished benchmarking commit (a2f1897): 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)	0.6%	[0.2%, 2.7%]	36
Regressions ❌ (secondary)	0.6%	[0.3%, 0.8%]	5
Improvements ✅ (primary)	-0.8%	[-0.8%, -0.7%]	4
Improvements ✅ (secondary)	-0.7%	[-0.7%, -0.6%]	3
All ❌✅ (primary)	0.4%	[-0.8%, 2.7%]	40

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)	2.4%	[0.8%, 5.0%]	9
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-5.1%	[-5.1%, -5.1%]	1
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)	1.5%	[1.5%, 1.5%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.5%	[1.5%, 1.5%]	1

Binary size

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

Bootstrap: 623.65s -> 622.591s (-0.17%)
Artifact size: 271.98 MiB -> 271.98 MiB (-0.00%)

Compute NLL loan scopes using the polonius model For a *location-insensitive* analysis (that is, without expressiveness improvements for users yet), this PR implements loans going out of scope using reachability and liveness, rather than checking if the issuing region's values contain a given CFG point. This is equivalent to NLL scopes and computes the same data. r? `@matthewjasper` A couple of notes: - there are some assumptions about SCC representatives, placeholders, free regions, and member constraints that I believe hold, and they're documented in the code - this passes all the UI tests with `-Zpolonius=next` -- the perf is [not terrible](rust-lang#112432 (comment)) and there are a bunch of ways to improve it in the future. - there's a fixme left, hopefully Matthew you know a clean way to get the information it mentions.

Compute NLL loan scopes using the polonius model For a *location-insensitive* analysis (that is, without expressiveness improvements for users yet), this PR implements loans going out of scope using reachability and liveness, rather than checking if the issuing region's values contain a given CFG point. This is equivalent to NLL scopes and computes the same data. r? `@matthewjasper` A couple of notes: - there are some assumptions about SCC representatives, placeholders, free regions, and member constraints that I believe hold, and they're documented in the code - this passes all the UI tests with `-Zpolonius=next` -- the perf is [not terrible](rust-lang/rust#112432 (comment)) and there are a bunch of ways to improve it in the future. - there's a fixme left, hopefully Matthew you know a clean way to get the information it mentions.

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Jun 8, 2023

This comment has been minimized.

Sign in to view

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Jun 8, 2023

This comment has been minimized.

Sign in to view

rustbot added perf-regression Performance regression. and removed S-waiting-on-perf Status: Waiting on a perf run to be completed. labels Jun 8, 2023

lqd closed this Jun 8, 2023

lqd reopened this Jun 9, 2023

lqd force-pushed the loan-liveness-frontier branch from fff2a32 to db6a863 Compare June 9, 2023 18:24

This comment has been minimized.

Sign in to view

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Jun 9, 2023

This comment has been minimized.

Sign in to view

rustbot removed the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Jun 9, 2023

lqd closed this Jun 9, 2023

lqd reopened this Jun 23, 2023

lqd force-pushed the loan-liveness-frontier branch from db6a863 to 6c85c49 Compare June 23, 2023 23:53

This comment has been minimized.

Sign in to view

lqd closed this Sep 8, 2023

lqd added 7 commits October 5, 2023 16:32

introduce Polonius enum for -Zpolonius

64d744d

this allows to opt into using the legacy version or the in-tree prototype

rename ancient regioncx

eb16d94

document Borrows dataflow analysis

4a9f830

extend Polonius options helpers

3dee8c9

add note why a function is pub

07bdefe

allow access to liveness' PointIndexes

10d37bb

compute NLL loan scopes with liveness in -Zpolonius=next

0f2156c

lqd reopened this Oct 5, 2023

lqd force-pushed the loan-liveness-frontier branch from da792f0 to 99e0fc2 Compare October 5, 2023 16:38

This comment has been minimized.

Sign in to view

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Oct 5, 2023

tmp: use polonius loan scopes to see how terribly slow it is

8d5cbc6

lqd force-pushed the loan-liveness-frontier branch from 99e0fc2 to 8d5cbc6 Compare October 5, 2023 17:07

This comment has been minimized.

Sign in to view

bors mentioned this pull request Oct 5, 2023

Remove is global hack #116417

Merged

This comment has been minimized.

Sign in to view

rustbot removed the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Oct 5, 2023

lqd closed this Oct 6, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[perf] Loan liveness frontier slowness evaluation #112432

[perf] Loan liveness frontier slowness evaluation #112432

lqd commented Jun 8, 2023

lqd commented Jun 8, 2023

This comment has been minimized.

bors commented Jun 8, 2023

bors commented Jun 8, 2023

This comment has been minimized.

rust-timer commented Jun 8, 2023

lqd commented Jun 8, 2023

lqd commented Jun 9, 2023

This comment has been minimized.

bors commented Jun 9, 2023

bors commented Jun 9, 2023

This comment has been minimized.

rust-timer commented Jun 9, 2023

lqd commented Jun 24, 2023

This comment has been minimized.

lqd commented Oct 5, 2023

This comment has been minimized.

bors commented Oct 5, 2023

lqd commented Oct 5, 2023

This comment has been minimized.

bors commented Oct 5, 2023

bors commented Oct 5, 2023

This comment has been minimized.

rust-timer commented Oct 5, 2023

[perf] Loan liveness frontier slowness evaluation #112432

[perf] Loan liveness frontier slowness evaluation #112432

Conversation

lqd commented Jun 8, 2023

lqd commented Jun 8, 2023

This comment has been minimized.

bors commented Jun 8, 2023

bors commented Jun 8, 2023

This comment has been minimized.

rust-timer commented Jun 8, 2023

Overall result: ❌ regressions - ACTION NEEDED

lqd commented Jun 8, 2023

lqd commented Jun 9, 2023

This comment has been minimized.

bors commented Jun 9, 2023

bors commented Jun 9, 2023

This comment has been minimized.

rust-timer commented Jun 9, 2023

Overall result: ❌✅ regressions and improvements - ACTION NEEDED

lqd commented Jun 24, 2023

This comment has been minimized.

lqd commented Oct 5, 2023

This comment has been minimized.

bors commented Oct 5, 2023

lqd commented Oct 5, 2023

This comment has been minimized.

bors commented Oct 5, 2023

bors commented Oct 5, 2023

This comment has been minimized.

rust-timer commented Oct 5, 2023

Overall result: ❌✅ regressions and improvements - ACTION NEEDED