Use `iter::repeat_n` to implement `Vec::extend_with` #133662

paolobarbolini · 2024-11-30T06:22:29Z

This replaces the Vec::extend_with manual implementation with iter::repeat_n, simplifying the code and fixing the issue identified in #120050.

rustbot · 2024-11-30T06:22:37Z

r? @Noratrieb

rustbot has assigned @Noratrieb.
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

joboet · 2024-11-30T12:47:40Z

Let's see what effect this has in the compiler!
@bors try @rust-timer queue

bors · 2024-11-30T12:48:51Z

⌛ Trying commit 4bf7072 with merge 243c5cc...

…epeatn, r=<try> Use `iter::repeat_n` to implement `Vec::extend_with` This replaces the `Vec::extend_with` manual implementation, which is used by `Vec::resize` and `Vec` `SpecFromElem`, with `iter::repeat_n`. I've compared the codegen output between: 1. the current `Vec::resize` impl 2. this branch 3. this branch + rust-lang#130887 3 gives the closest codegen output to 1, with some output improvements. 2 doesn't look good: https://rust.godbolt.org/z/Yrc83EhjY. May also help rust-lang#120050?

Noratrieb · 2024-11-30T13:48:12Z

given that this branch doesn't look good without #130887, let's block it on that.
@rustbot blocked

bors · 2024-11-30T14:37:20Z

☀️ Try build successful - checks-actions
Build commit: 243c5cc (243c5cc6905efbdf778394d96f4b83701962964f)

rust-timer · 2024-11-30T17:09:58Z

Finished benchmarking commit (243c5cc): comparison URL.

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

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 the most reliable metric that we have; it was used to determine the overall result at the top of this comment. However, even this metric can sometimes exhibit noise.

	mean	range	count
Regressions ❌ (primary)	0.3%	[0.1%, 0.7%]	16
Regressions ❌ (secondary)	0.4%	[0.1%, 0.6%]	9
Improvements ✅ (primary)	-0.9%	[-0.9%, -0.9%]	1
Improvements ✅ (secondary)	-1.3%	[-1.8%, -0.5%]	8
All ❌✅ (primary)	0.3%	[-0.9%, 0.7%]	17

Max RSS (memory usage)

Results (primary 0.1%)

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%	[0.9%, 11.4%]	3
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-8.1%	[-11.1%, -5.0%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.1%	[-11.1%, 11.4%]	5

Cycles

Results (secondary -2.1%)

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)	-2.1%	[-2.1%, -2.1%]	1
All ❌✅ (primary)	-	-	0

Binary size

Results (primary 0.4%)

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)	0.5%	[0.0%, 1.3%]	43
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.1%	[-0.2%, -0.0%]	7
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.4%	[-0.2%, 1.3%]	50

Bootstrap: 773.805s -> 774.205s (0.05%)
Artifact size: 332.32 MiB -> 331.64 MiB (-0.20%)

paolobarbolini · 2024-12-01T09:37:08Z

Could we do another perf run? I've cherry-picked the commit from #130887 since, at least via Compiler Explorer, the codegen looked better

joboet · 2024-12-01T11:04:19Z

Sure!
@bors try @rust-timer queue

bors · 2024-12-01T11:05:32Z

⌛ Trying commit b080e18 with merge 78d5ed4...

…epeatn, r=<try> Use `iter::repeat_n` to implement `Vec::extend_with` This replaces the `Vec::extend_with` manual implementation, which is used by `Vec::resize` and `Vec` `SpecFromElem`, with `iter::repeat_n`. I've compared the codegen output between: 1. the current `Vec::resize` impl 2. this branch 3. this branch + rust-lang#130887 3 gives the closest codegen output to 1, with some output improvements. 2 doesn't look good: https://rust.godbolt.org/z/Yrc83EhjY. May also help rust-lang#120050? --- WARNING: DO NOT MERGE - in order to run the perf run in rust-lang#133662 (comment) this PR currently also contains commits from rust-lang#130887

bors · 2024-12-01T12:50:21Z

☀️ Try build successful - checks-actions
Build commit: 78d5ed4 (78d5ed40d7c459a7e5ea4cb6bea9918a493a800f)

rust-timer · 2024-12-01T14:18:32Z

Finished benchmarking commit (78d5ed4): comparison URL.

Overall result: ❌ regressions - please read the text below

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 the most reliable metric that we have; it was used to determine the overall result at the top of this comment. However, even this metric can sometimes exhibit noise.

	mean	range	count
Regressions ❌ (primary)	0.4%	[0.1%, 0.7%]	17
Regressions ❌ (secondary)	1.3%	[0.1%, 3.2%]	16
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.4%	[0.1%, 0.7%]	17

Max RSS (memory usage)

Results (primary -5.6%, secondary 0.3%)

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)	5.3%	[5.3%, 5.3%]	1
Regressions ❌ (secondary)	1.5%	[1.0%, 1.8%]	3
Improvements ✅ (primary)	-11.1%	[-15.5%, -6.8%]	2
Improvements ✅ (secondary)	-3.2%	[-3.2%, -3.2%]	1
All ❌✅ (primary)	-5.6%	[-15.5%, 5.3%]	3

Cycles

Results (primary -1.5%, secondary 35.1%)

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)	0.9%	[0.9%, 0.9%]	1
Regressions ❌ (secondary)	35.1%	[33.0%, 39.1%]	6
Improvements ✅ (primary)	-4.0%	[-4.0%, -4.0%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-1.5%	[-4.0%, 0.9%]	2

Binary size

Results (primary 0.6%)

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)	0.7%	[0.1%, 1.7%]	33
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-0.1%	[-0.2%, -0.0%]	9
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.6%	[-0.2%, 1.7%]	42

Bootstrap: 767.632s -> 769.266s (0.21%)
Artifact size: 332.19 MiB -> 332.16 MiB (-0.01%)

paolobarbolini · 2024-12-02T09:12:17Z

Looking at performance data borrow checking time in the Cycles benchmark for wg-grammar is the main culprit. I'm confused why that is.

The fact that this also fixes #120050 makes these changes still worth it to me.

Soveu · 2025-06-18T20:29:46Z

@paolobarbolini #130887 has been merged
@rustbot label -S-blocked +S-waiting-on-author

paolobarbolini · 2025-06-18T21:04:29Z

I've rebased on top of master. Should we do another perf run?

scottmcm · 2025-06-19T04:46:59Z

Sure, with the rebase we might as well give it a shot

@bors try @rust-timer queue

…=<try> Use `iter::repeat_n` to implement `Vec::extend_with` This replaces the `Vec::extend_with` manual implementation with `iter::repeat_n`, simplifying the code and also possibly fixing the issue identified in #120050.

bors · 2025-06-19T04:48:11Z

⌛ Trying commit 29565d7 with merge c79c24a...

bors · 2025-06-19T07:18:51Z

☀️ Try build successful - checks-actions
Build commit: c79c24a (c79c24ad439e510f1baa231dcaf131000f552308)

rust-timer · 2025-06-19T12:05:29Z

Finished benchmarking commit (c79c24a): 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)	0.7%	[0.2%, 3.0%]	12
Regressions ❌ (secondary)	3.0%	[0.2%, 7.5%]	16
Improvements ✅ (primary)	-0.9%	[-1.3%, -0.5%]	2
Improvements ✅ (secondary)	-0.5%	[-1.2%, -0.2%]	3
All ❌✅ (primary)	0.5%	[-1.3%, 3.0%]	14

Max RSS (memory usage)

Results (primary 2.2%, secondary -0.1%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	4.9%	[1.1%, 7.3%]	4
Regressions ❌ (secondary)	4.7%	[3.3%, 6.5%]	3
Improvements ✅ (primary)	-3.3%	[-4.2%, -2.4%]	2
Improvements ✅ (secondary)	-7.2%	[-8.5%, -5.9%]	2
All ❌✅ (primary)	2.2%	[-4.2%, 7.3%]	6

Cycles

Results (primary 1.7%, secondary 3.6%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	1.7%	[0.7%, 3.6%]	3
Regressions ❌ (secondary)	3.6%	[2.2%, 10.5%]	7
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.7%	[0.7%, 3.6%]	3

Binary size

Results (primary 0.6%, secondary 0.1%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	0.6%	[0.1%, 1.7%]	38
Regressions ❌ (secondary)	0.1%	[0.1%, 0.1%]	1
Improvements ✅ (primary)	-0.1%	[-0.1%, -0.1%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.6%	[-0.1%, 1.7%]	39

Bootstrap: 692.944s -> 693.357s (0.06%)
Artifact size: 372.01 MiB -> 371.96 MiB (-0.01%)

paolobarbolini · 2025-07-25T08:55:09Z

@scottmcm Do you think that these results are worth it given that this also acts as an alternative to #120050?

scottmcm · 2025-07-25T16:53:30Z

Personally I still want to do this, but at the same time with my reviewer hat on I don't know if I can justify the perf numbers. Anywhere outside the standard library I'd just do it, but Vec is important enough that we'll take less-elegant implementations for perf reasons :/

I wonder if maybe there's a tweak for it that could help. An idea that came to mind: what if it specialized on Copy, with that path having the obvious loop (which wouldn't even need SetLenOnDrop), and used repeat_n for the more-complex non-Copy path. My instinct is that the image regression is probably dealing in trivial Copy stuff, so emitting simpler code for that (no we-have-defer-at-home codepath) could even be good.

EDIT: or, I suppose, if we're not allowed to specialize on Copy any more this could just branch on if const { needs_drop::<T>() }? I don't know that that would help as much, though.

This replaces the manual `Vec::extend_with` implementation with `iter::repeat_n` and `Vec::extend_trusted`. This simplifies the code and gets LLVM to remove the special case for the last element when `T` is trivial to clone.

After running perf on the previous commit, it was found that the compile-time regression was too significant to ignore. This introduces a specialized, simplified implementation for `T: Copy` that will contain the perf regression.

paolobarbolini · 2025-07-26T03:45:20Z

Specializing on Copy sounds like a great idea. This should now be ready for another perf run.

scottmcm · 2025-07-26T04:02:15Z

Let's give it a shot!

@bors try @rust-timer queue

rust-bors · 2025-07-26T04:02:19Z

⌛ Trying commit bbcb772 with merge 4f5a239…

To cancel the try build, run the command @bors try cancel.

…=<try> Use `iter::repeat_n` to implement `Vec::extend_with`

Soveu · 2025-07-26T04:10:53Z

library/alloc/src/vec/spec_extend_with.rs

+        // SAFETY: the above `reserve` call guarantees `n` to be in bounds.
+        let unfilled = unsafe { unfilled.get_unchecked_mut(..n) };
+        for elem in unfilled {
+            *elem = MaybeUninit::new(value);


Wouldn't unfilled.fill(MaybeUninit::new(value)) fit here better?

Lots of options. Because it's copy could even just *p.add(i) = value;.

But let's see what the perf run says before worrying too much about it.

rust-bors · 2025-07-26T06:17:39Z

☀️ Try build successful (CI)
Build commit: 4f5a239 (4f5a23979e54bf7cdad9d71f114d5e3ccbf1d3cf, parent: 0dd07bd2165db70ffbe0b903635de26ca630bf20)

rust-timer · 2025-07-26T08:23:22Z

Finished benchmarking commit (4f5a239): 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)	0.2%	[0.1%, 0.7%]	10
Regressions ❌ (secondary)	2.1%	[0.1%, 6.8%]	19
Improvements ✅ (primary)	-0.2%	[-0.7%, -0.1%]	26
Improvements ✅ (secondary)	-0.3%	[-0.6%, -0.2%]	16
All ❌✅ (primary)	-0.1%	[-0.7%, 0.7%]	36

Max RSS (memory usage)

Results (primary 1.0%, secondary -2.0%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	2.7%	[1.5%, 4.0%]	3
Regressions ❌ (secondary)	2.0%	[0.8%, 3.2%]	2
Improvements ✅ (primary)	-3.9%	[-3.9%, -3.9%]	1
Improvements ✅ (secondary)	-4.6%	[-6.5%, -2.6%]	3
All ❌✅ (primary)	1.0%	[-3.9%, 4.0%]	4

Cycles

Results (secondary 4.0%)

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)	4.0%	[2.4%, 5.3%]	6
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Binary size

Results (primary 0.4%, secondary 0.3%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	0.5%	[0.1%, 1.1%]	30
Regressions ❌ (secondary)	0.3%	[0.3%, 0.3%]	1
Improvements ✅ (primary)	-0.2%	[-0.2%, -0.2%]	2
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.4%	[-0.2%, 1.1%]	32

Bootstrap: 467.876s -> 469.154s (0.27%)
Artifact size: 376.61 MiB -> 376.65 MiB (0.01%)

rustbot assigned Noratrieb Nov 30, 2024

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Nov 30, 2024