Skip to content

Compilation time regression in 1.73 when building the fundsp crate #116797

New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

#

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

Jump to bottom
Closed
irh opened this issue Oct 16, 2023 · 6 comments
Closed

Compilation time regression in 1.73 when building the fundsp crate #116797

irh opened this issue Oct 16, 2023 · 6 comments
Assignees
@nnethercote
Labels
C-bug Category: This is a bug. I-compiletime Issue: Problems and improvements with respect to compile times. P-high High priority regression-from-stable-to-stable Performance or correctness regression from one stable version to another. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue.

Comments

@irh
Copy link

irh commented Oct 16, 2023

1.73 introduced a regression in compilation time when building the fundsp crate.

Previously, I could expect a clean build of the crate to take ~22s. Now, compilation hangs for several minutes, I haven't left it running longer than that.

I bisected to #114611, and 8378487 appears to be the commit that introduces the slowdown. Reverting the commit on the latest master brings fundsp build times back to normal.
@irh irh added C-bug Category: This is a bug. regression-untriaged Untriaged performance or correctness regression. labels Oct 16, 2023
@rustbot rustbot added I-prioritize Issue: Indicates that prioritization has been requested for this issue. needs-triage This issue may need triage. Remove it if it has been sufficiently triaged. regression-from-stable-to-stable Performance or correctness regression from one stable version to another. and removed regression-untriaged Untriaged performance or correctness regression. labels Oct 16, 2023
@saethlin saethlin added I-compiletime Issue: Problems and improvements with respect to compile times. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. and removed needs-triage This issue may need triage. Remove it if it has been sufficiently triaged. labels Oct 16, 2023
@saethlin
Copy link
Member

I think this is the same root cause as #116624 and #116780.

@Noratrieb
Copy link
Member

@nnethercote

@nnethercote nnethercote self-assigned this Oct 16, 2023
@nnethercote
Copy link
Contributor

nnethercote commented Oct 17, 2023 •
edited
Loading

I think this is the same root cause as #116624 and #116780.

It is the same as #116780. It is not the same as #116624, which is slow even in 1.72.

nnethercote added a commit to nnethercote/rust that referenced this issue Oct 17, 2023
@nnethercote
Fix a performance regression in obligation deduplication.
91f2fbc 
Commit 8378487 from rust-lang#114611 changed the location of an obligation
deduplication step in `opt_normalize_projection_type`. This meant that
deduplication stopped happening on one path where it was still
necessary, causing a couple of drastic performance regressions.

This commit moves the deduplication back to the old location. The good
news is that rust-lang#114611 had four commits and 8378487 was of minimal
importance, so the perf benefits from that PR remain.

Fixes rust-lang#116780, rust-lang#116797.
@nnethercote nnethercote mentioned this issue Oct 17, 2023
Merged
@nnethercote
Copy link
Contributor

Good news: #116826 will fix the regression.

Even better news: on my machine this crate took 3m13s to compile. After removing the 3rd and 4th commits from #114611, it dropped to 22s. After reinstating the 4th commit (because the 3rd commit was the problem) it dropped to 9s :)

bors added a commit to rust-lang-ci/rust that referenced this issue Oct 17, 2023
@bors
Auto merge of rust-lang#116826 - nnethercote:fix-116780-116797, r=com…
39164b8 
…piler-errors

Fix a performance regression in obligation deduplication.

Commit 8378487 from rust-lang#114611 changed the location of an obligation deduplication step in `opt_normalize_projection_type`. This meant that deduplication stopped happening on one path where it was still necessary, causing a couple of drastic performance regressions.

This commit moves the deduplication back to the old location. The good news is that rust-lang#114611 had four commits and 8378487 was of minimal importance, so the perf benefits from that PR remain.

Fixes rust-lang#116780, rust-lang#116797.

r? `@compiler-errors`
@apiraino
Copy link
Contributor

WG-prioritization assigning priority (Zulip discussion).

@rustbot label -I-prioritize +P-high

@irh in the next day a new nightly with a fix should be out. If you have a chance, feel free to test it and confirm if it fixes for you too. thanks for reporting the issue.

@rustbot rustbot added P-high High priority and removed I-prioritize Issue: Indicates that prioritization has been requested for this issue. labels Oct 17, 2023
@irh
Copy link
Author

irh commented Oct 17, 2023

@apiraino I tested it on master, and can confirm it's fixed.

@nnethercote Thanks for taking a look at this so quickly! I also get the 9s build time, very nice 🥳

@irh irh closed this as completed Oct 17, 2023
cuviper pushed a commit to cuviper/rust that referenced this issue Oct 21, 2023
@nnethercote @cuviper
Fix a performance regression in obligation deduplication.
e45b2ab 
Commit 8378487 from rust-lang#114611 changed the location of an obligation
deduplication step in `opt_normalize_projection_type`. This meant that
deduplication stopped happening on one path where it was still
necessary, causing a couple of drastic performance regressions.

This commit moves the deduplication back to the old location. The good
news is that rust-lang#114611 had four commits and 8378487 was of minimal
importance, so the perf benefits from that PR remain.

Fixes rust-lang#116780, rust-lang#116797.

(cherry picked from commit 91f2fbc)
xobs pushed a commit to betrusted-io/rust that referenced this issue Nov 7, 2023
@nnethercote @xobs
Fix a performance regression in obligation deduplication.
4c448dc 
Commit 8378487 from rust-lang#114611 changed the location of an obligation
deduplication step in `opt_normalize_projection_type`. This meant that
deduplication stopped happening on one path where it was still
necessary, causing a couple of drastic performance regressions.

This commit moves the deduplication back to the old location. The good
news is that rust-lang#114611 had four commits and 8378487 was of minimal
importance, so the perf benefits from that PR remain.

Fixes rust-lang#116780, rust-lang#116797.

(cherry picked from commit 91f2fbc)
# for free to join this conversation on GitHub. Already have an account? # to comment
Assignees

@nnethercote nnethercote

Labels
C-bug Category: This is a bug. I-compiletime Issue: Problems and improvements with respect to compile times. P-high High priority regression-from-stable-to-stable Performance or correctness regression from one stable version to another. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue.
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
6 participants
@irh @nnethercote @apiraino @saethlin @rustbot @Noratrieb