[TaKO8Ki]

I used a similar approach for DefId encoding and decoding, based on the existing Symbol encoder and decoder.

rust/compiler/rustc_middle/src/query/on_disk_cache.rs

Line 818 in 8190021

fn encode_symbol_or_byte_symbol(

rust/compiler/rustc_middle/src/query/on_disk_cache.rs

Line 447 in 8190021

fn decode_symbol_or_byte_symbol<S>(

[TaKO8Ki]

@bors try @rust-timer queue

[rust-bors]

☀️ Try build successful (CI)
Build commit: 381214b (381214b82df0417c4443eb8d1e89530cdc28c058)
Base parent: 2d37576 (2d37576c3d7134aa2d5c2f2396712a8b4a38ef6f)

[rust-timer]

Finished benchmarking commit (381214b): comparison URL.

Overall result: ❌ regressions - please read:

Benchmarking means the PR may be perf-sensitive. It's automatically marked not fit for rolling up. Overriding is possible but disadvised: it risks changing compiler perf.

Next, please: If you can, justify the regressions found in this try perf run in writing along with @rustbot label: +perf-regression-triaged. If not, fix the regressions and do another perf run. Neutral or positive results will clear the label automatically.

@bors rollup=never rustc-perf
@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)	0.8%	[0.1%, 1.5%]	132
Regressions ❌ (secondary)	0.7%	[0.1%, 2.1%]	89
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.8%	[0.1%, 1.5%]	132

Max RSS (memory usage)

Results (primary -1.7%, secondary -4.2%)

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)	-	-	0
Improvements ✅ (primary)	-1.7%	[-2.9%, -1.0%]	29
Improvements ✅ (secondary)	-4.2%	[-5.3%, -1.5%]	7
All ❌✅ (primary)	-1.7%	[-2.9%, -1.0%]	29

Cycles

Results (secondary -1.1%)

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)	2.0%	[2.0%, 2.0%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-4.1%	[-4.1%, -4.1%]	1
All ❌✅ (primary)	-	-	0

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 487.613s -> 488.634s (0.21%)
Artifact size: 406.81 MiB -> 406.90 MiB (0.02%)

[TaKO8Ki]

Instruction-count regressions were concentrated in incremental rebuilds, averaging 0.8% in the affected primary cases and 0.7% in the affected secondary cases.

Query-cache size decreased by 25.71% on average across the relevant primary and secondary cases.

https://perf.rust-lang.org/compare.html?start=2d37576c3d7134aa2d5c2f2396712a8b4a38ef6f&end=381214b82df0417c4443eb8d1e89530cdc28c058&stat=size%3Aquery_cache&tab=compile

[Kobzol]

Nice wins there. I tried building (cargo build) the https://github.com/It4innovations/hyperqueue crate before/after this PR. Size of the target dir:

• Before: 1527784 B
• After: 1501924 B

So about a 2% (~25 MiB) reduction here, for a clean incremental build.

[rustbot]

r? @beetrees

rustbot has assigned @beetrees.
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

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: compiler, query-system
• compiler, query-system expanded to 77 candidates
• Random selection from 18 candidates

[panstromek]

Most of the icount increase in serde opt incr-unchanged comes from Span encoding and decoding. That seems a bit weird. Is there something unexpected going on? How does this interact with spans?

30,275,316 (100.0%)  PROGRAM TOTALS

--------------------------------------------------------------------------------
-- File:function summary
--------------------------------------------------------------------------------
  Ir________________________  file:function

< 29,982,329 (99.0%,  99.0%)  ???:
  11,173,357 (36.9%)            <rustc_middle::query::on_disk_cache::CacheEncoder as rustc_span::SpanEncoder>::encode_span
   9,874,207 (32.6%)            <rustc_middle::query::on_disk_cache::CacheDecoder as rustc_span::SpanDecoder>::decode_span
   3,952,030 (13.1%)            <rustc_middle::query::on_disk_cache::CacheDecoder as rustc_span::SpanDecoder>::decode_def_id
   3,616,469 (11.9%)            <rustc_middle::query::on_disk_cache::CacheEncoder as rustc_span::SpanEncoder>::encode_def_id
   1,228,968  (4.1%)            <rustc_middle::mir::syntax::Operand as rustc_serialize::serialize::Decodable<rustc_middle::query::on_disk_cache::CacheDecoder>>::decode
    -867,459 (-2.9%)            <rustc_middle::mir::syntax::Place as rustc_serialize::serialize::Decodable<rustc_middle::query::on_disk_cache::CacheDecoder>>::decode
    -861,145 (-2.8%)            <alloc::vec::Vec<rustc_span::Spanned<rustc_middle::mir::syntax::Operand>> as rustc_serialize::serialize::Decodable<rustc_middle::query::on_disk_cache::CacheDecoder>>::decode
     836,200  (2.8%)            <rustc_middle::query::on_disk_cache::CacheDecoder as rustc_span::BlobDecoder>::decode_local_def_id
     704,292  (2.3%)            rustc_query_impl::incremental::encode_query_values_inner::<rustc_middle::query::caches::DefIdCache<rustc_middle::query::erase::ErasedData<[u8; 8]>>, &rustc_middle::ty::generics::Generics>::{closure#0}
     567,527  (1.9%)            <hashbrown::raw::RawTable<(rustc_middle::ty::Ty, usize)>>::reserve_rehash::<hashbrown::map::make_hasher<rustc_middle::ty::Ty, usize, rustc_hash::FxBuildHasher>::{closure#0}>
    -537,260 (-1.8%)            <rustc_stable_hash::sip128::SipHasher128>::finish128_inner

[TaKO8Ki]

@panstromek SpanData contains a parent Option<LocalDefId>, and this change also affects how LocalDefId is encoded and decoded. I think this is contributing to the increased cost of Span encoding and decoding.

rust/compiler/rustc_middle/src/query/on_disk_cache.rs

Line 865 in 8190021

span_data.parent.encode(self);

[beetrees]

@bors r+

[rust-bors]

📌 Commit 8190021 has been approved by beetrees

It is now in the queue for this repository.

🌲 The tree is currently closed for pull requests below priority 5. This pull request will be tested once the tree is reopened.

Reason for tree closure: test-aarch64-apple-macos-26-2 out of disk space

[JonathanBrouwer]

@bors rollup=iffy
Adding this to a perf-sensitive rollup

[rust-timer]

Note

This PR was benchmarked as part of triage of its containing rollup: triage URL.

Finished benchmarking commit (69cf161): comparison URL.

Overall result: ❌ regressions - please read:

Our benchmarks found a performance regression caused by this PR.
This might be an actual regression, but it can also be just noise.

Next Steps:

• If the regression was expected or you think it can be justified,
  please write a comment with sufficient written justification, and add
  @rustbot label: +perf-regression-triaged to it, to mark the regression as triaged.
• If you think that you know of a way to resolve the regression, try to create
  a new PR with a fix for the regression.
• If you do not understand the regression or you think that it is just noise,
  you can ask the @rust-lang/wg-compiler-performance working group for help (members of this group
  were already notified of this PR).

@rustbot label: +perf-regression
cc @rust-lang/wg-compiler-performance

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)	0.8%	[0.1%, 1.6%]	149
Regressions ❌ (secondary)	0.7%	[0.1%, 2.2%]	81
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.8%	[0.1%, 1.6%]	149

Max RSS (memory usage)

Results (primary -1.1%, 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)	0.9%	[0.4%, 1.8%]	10
Regressions ❌ (secondary)	0.8%	[0.4%, 1.6%]	4
Improvements ✅ (primary)	-1.7%	[-6.5%, -0.5%]	37
Improvements ✅ (secondary)	-3.4%	[-5.2%, -0.9%]	10
All ❌✅ (primary)	-1.1%	[-6.5%, 1.8%]	47

Cycles

Results (primary 0.0%, secondary 0.8%)

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

	mean	range	count
Regressions ❌ (primary)	0.8%	[0.5%, 1.2%]	8
Regressions ❌ (secondary)	1.5%	[1.0%, 2.4%]	4
Improvements ✅ (primary)	-0.9%	[-2.2%, -0.4%]	7
Improvements ✅ (secondary)	-0.6%	[-0.7%, -0.5%]	2
All ❌✅ (primary)	0.0%	[-2.2%, 1.2%]	15

Binary size

This perf run didn't have relevant results for this metric.

Artifact size: 405.91 MiB -> 405.98 MiB (0.02%)

[panstromek]

perf triage:

This impact was known before merge. This PR trades small icount regressions for large query cache size improvements.

@rustbot label: +perf-regression-triaged