Preprocess dominator tree to answer queries in O(1) #107157

tmiasko · 2023-01-21T13:22:10Z

rustbot · 2023-01-21T13:22:14Z

Some changes occurred to MIR optimizations

cc @rust-lang/wg-mir-opt

bors · 2023-01-21T15:19:44Z

☔ The latest upstream changes (presumably #106976) made this pull request unmergeable. Please resolve the merge conflicts.

There is a number of APIs that answer dominance queries. Previously they were named either "dominates" or "is_dominated_by". Consistently use the "dominates" form. No functional changes.

tmiasko · 2023-01-22T00:04:44Z

@bors try @rust-timer queue

bors · 2023-01-22T00:04:53Z

⌛ Trying commit fff5775 with merge 9c8bbf49ba0cc73e86c499b73a28c0b60b2af24c...

bors · 2023-01-22T02:49:11Z

☀️ Try build successful - checks-actions
Build commit: 9c8bbf49ba0cc73e86c499b73a28c0b60b2af24c (9c8bbf49ba0cc73e86c499b73a28c0b60b2af24c)

rust-timer · 2023-01-22T04:38:52Z

Finished benchmarking commit (9c8bbf49ba0cc73e86c499b73a28c0b60b2af24c): 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)	0.2%	[0.2%, 0.3%]	2
Regressions ❌ (secondary)	0.5%	[0.2%, 0.8%]	4
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.2%	[0.2%, 0.3%]	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)	-	-	0
Regressions ❌ (secondary)	2.4%	[2.4%, 2.4%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
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)	1.3%	[0.7%, 1.8%]	14
Regressions ❌ (secondary)	2.9%	[1.8%, 3.9%]	18
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.3%	[0.7%, 1.8%]	14

bryangarza

I'm not an official reviewer, but wanted to leave some feedback.

Just had one comment in the code itself, other than that, it'd be nice to see a unit test or two for the dominators function before merging.

bryangarza · 2023-01-24T21:49:41Z

compiler/rustc_data_structures/src/graph/dominators/mod.rs

+        e.start = m;
+        e.end = m;
+    }
+    let mut node = IndexVec::from_elem_n(Idx::new(0), m.try_into().unwrap());


if you make the m a usize from the start, you can avoid the try_into().unwrap()

pnkfelix · 2023-02-23T15:49:56Z

@tmiasko you wrote in a separate comment: #107449 (comment)

FYI: #107157 is intended for instrument coverage builds, where the cost of dominance queries often exceeds the cost of initial preprocessing - the opposite of a profile here, where at best it wouldn't be a regression.

I'm trying to connect that statement with the PR here.

Is the preprocessing done here supposed to be conditionally enabled? I haven't yet seen that in this PR, but I only did a quick skim so maybe I overlooked it.

tmiasko · 2023-02-24T23:37:43Z

Is the preprocessing done here supposed to be conditionally enabled?

The extra cost seems negligible, so it can be unconditional (one can't quite tell from last perf results alone, since the measurement is biased by a regression in obligation processing, which has unstable codegen for some reason).

This pull request isn't quite ready for review (hence no assigned reviewer). I also wanted to add a new benchmark that demonstrates perf aspect addressed here first.

Closing for now, maybe will return to this later.

…illot,tmiasko Preprocess and cache dominator tree Preprocessing dominators has a very strong effect for rust-lang#111344. That pass checks that assignments dominate their uses repeatedly. Using the unprocessed dominator tree caused a quadratic runtime (number of bbs x depth of the dominator tree). This PR also caches the dominator tree and the pre-processed dominators in the MIR cfg cache. Rebase of rust-lang#107157 cc `@tmiasko`

…asko Preprocess and cache dominator tree Preprocessing dominators has a very strong effect for rust-lang/rust#111344. That pass checks that assignments dominate their uses repeatedly. Using the unprocessed dominator tree caused a quadratic runtime (number of bbs x depth of the dominator tree). This PR also caches the dominator tree and the pre-processed dominators in the MIR cfg cache. Rebase of rust-lang/rust#107157 cc `@tmiasko`

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 Jan 21, 2023

tmiasko added 2 commits January 22, 2023 00:24

Consistently use dominates instead of is_dominated_by

ab39f2d

There is a number of APIs that answer dominance queries. Previously they were named either "dominates" or "is_dominated_by". Consistently use the "dominates" form. No functional changes.

Preprocess dominator tree to answer queries in O(1)

fff5775

tmiasko force-pushed the dominators-O(1) branch from 01e1cc6 to fff5775 Compare January 21, 2023 23:25

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 Jan 22, 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 Jan 22, 2023

bryangarza suggested changes Jan 24, 2023

View reviewed changes

cjgillot mentioned this pull request Feb 12, 2023

Enable CopyProp #107449

Merged

tmiasko closed this Feb 24, 2023

cjgillot mentioned this pull request May 17, 2023

Preprocess and cache dominator tree #111673

Merged

tmiasko deleted the dominators-O(1) branch November 10, 2023 23:47

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Preprocess dominator tree to answer queries in O(1) #107157

Preprocess dominator tree to answer queries in O(1) #107157

tmiasko commented Jan 21, 2023

rustbot commented Jan 21, 2023

bors commented Jan 21, 2023

tmiasko commented Jan 22, 2023

This comment has been minimized.

bors commented Jan 22, 2023

bors commented Jan 22, 2023

This comment has been minimized.

rust-timer commented Jan 22, 2023

bryangarza left a comment

bryangarza Jan 24, 2023

pnkfelix commented Feb 23, 2023

tmiasko commented Feb 24, 2023

Preprocess dominator tree to answer queries in O(1) #107157

Preprocess dominator tree to answer queries in O(1) #107157

Conversation

tmiasko commented Jan 21, 2023

rustbot commented Jan 21, 2023

bors commented Jan 21, 2023

tmiasko commented Jan 22, 2023

This comment has been minimized.

bors commented Jan 22, 2023

bors commented Jan 22, 2023

This comment has been minimized.

rust-timer commented Jan 22, 2023

Overall result: ❌ regressions - ACTION NEEDED

Instruction count

Max RSS (memory usage)

Cycles

bryangarza left a comment

Choose a reason for hiding this comment

bryangarza Jan 24, 2023

Choose a reason for hiding this comment

pnkfelix commented Feb 23, 2023

tmiasko commented Feb 24, 2023