Preintern some TyKind::Bound values

[nnethercote]

The new trait solver produces a lot of these.

r? @compiler-errors

[nnethercote]

@bors try @rust-timer queue

[rust-bors]

⌛ Trying commit 837b610 with merge 06fbedc…

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

[rust-bors]

☀️ Try build successful (CI)
Build commit: 06fbedc (06fbedc70773f4b817a3a111c862cd846e4706c5, parent: b56aaec52bc0fa35591a872fb4aac81f606e265c)

[rust-timer]

Finished benchmarking commit (06fbedc): comparison URL.

Overall result: ✅ improvements - no action needed

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.

@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%	[0.2%, 0.2%]	1
Regressions ❌ (secondary)	0.2%	[0.2%, 0.2%]	1
Improvements ✅ (primary)	-0.3%	[-0.5%, -0.1%]	17
Improvements ✅ (secondary)	-0.7%	[-1.7%, -0.1%]	25
All ❌✅ (primary)	-0.3%	[-0.5%, 0.2%]	18

Max RSS (memory usage)

Results (primary 0.4%, secondary -2.6%)

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

	mean	range	count
Regressions ❌ (primary)	3.1%	[3.1%, 3.1%]	1
Regressions ❌ (secondary)	3.3%	[3.3%, 3.3%]	1
Improvements ✅ (primary)	-2.4%	[-2.4%, -2.4%]	1
Improvements ✅ (secondary)	-3.6%	[-6.7%, -1.2%]	6
All ❌✅ (primary)	0.4%	[-2.4%, 3.1%]	2

Cycles

Results (secondary 2.2%)

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)	2.2%	[0.8%, 3.5%]	2
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Binary size

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

Bootstrap: 469.891s -> 470.107s (0.05%)
Artifact size: 374.63 MiB -> 374.66 MiB (0.01%)

[lqd]

(note about results: I saw in other PRs that a handful of benchmarks are currently bimodal)

[nnethercote]

The new-solver benchmark wins are all real, they are the target of the change and I saw them locally:

The new trait solver produces a lot more Bound types. They are mostly produced by the call to Ty::new_anon_bound at the bottom of Canonicalizer::cached_fold_ty. Could there be a deeper optimization that would avoid the need to create these types at all, rather than making them cheaper to create? That would be nice...

[rustbot]

compiler-errors is not on the review rotation at the moment.
They may take a while to respond.

[nnethercote]

r? @lcnr

[compiler-errors]

Is this still a draft? I think that this is fine for now.

[compiler-errors]

Could there be a deeper optimization that would avoid the need to create these types at all, rather than making them cheaper to create? That would be nice...

This seems pretty difficult and probably not worth blocking this on that.

[nnethercote]

Correct, this no longer needs to be a draft.

This seems pretty difficult and probably not worth blocking this on that.

Ok. Any time I do a shallow "make operation X faster" micro-optimization I always wonder if there is a deeper and better "avoid operation X altogether" change that could be done instead.