Store a index per dep node kind

[Zoxc]

This stores an index map per dep node kind instead of using a single index map. This avoids storing DepKind in the key for greater cache efficiency.

Benchmark	Before	After		Before	After
	Time	Time	%	Memory	Memory	%
🟣 clap:check:unchanged	0.4376s	0.4334s	-0.95%	92.53 MiB	92.39 MiB	-0.15%
🟣 hyper:check:unchanged	0.1442s	0.1446s	0.29%	47.48 MiB	47.63 MiB	0.31%
🟣 regex:check:unchanged	0.3296s	0.3261s	💚 -1.06%	72.37 MiB	71.20 MiB	💚 -1.62%
🟣 syn:check:unchanged	0.6149s	0.6014s	💚 -2.19%	105.74 MiB	101.69 MiB	💚 -3.83%
🟣 syntex_syntax:check:unchanged	1.4882s	1.4560s	💚 -2.16%	209.75 MiB	201.46 MiB	💚 -3.95%
Total	3.0144s	2.9616s	💚 -1.75%	527.87 MiB	514.37 MiB	💚 -2.56%
Summary	1.0000s	0.9879s	💚 -1.21%	1 byte	0.98 bytes	💚 -1.85%

[rustbot]

r? @b-naber

(rustbot has picked a reviewer for you, use r? to override)

[lqd]

@bors try @rust-timer queue

[bors]

⌛ Trying commit 769a81e with merge e190874...

[bors]

☀️ Try build successful - checks-actions
Build commit: e190874 (e1908748729226ac01c251cd44097a9bafb5cbfe)

[rust-timer]

Finished benchmarking commit (e190874): 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)	1.3%	[0.3%, 2.3%]	104
Regressions ❌ (secondary)	1.5%	[0.4%, 2.2%]	42
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	1.3%	[0.3%, 2.3%]	104

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)	4.1%	[4.0%, 4.3%]	2
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-2.6%	[-5.3%, -1.0%]	19
Improvements ✅ (secondary)	-4.3%	[-4.4%, -4.3%]	3
All ❌✅ (primary)	-2.0%	[-5.3%, 4.3%]	21

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)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-3.2%	[-6.6%, -1.5%]	70
Improvements ✅ (secondary)	-5.6%	[-10.8%, -2.0%]	18
All ❌✅ (primary)	-3.2%	[-6.6%, -1.5%]	70

Binary size

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

Bootstrap: 630.365s -> 630.015s (-0.06%)
Artifact size: 318.00 MiB -> 317.98 MiB (-0.01%)

[Zoxc]

I included a commit where dep kind counts are also encoded so the index hash maps can get the right initial capacity. That's an additional improvement:

Benchmark	Before	After		Before	After
	Time	Time	%	Memory	Memory	%
🟣 clap:check:unchanged	0.4117s	0.4096s	-0.49%	93.09 MiB	92.73 MiB	-0.39%
🟣 hyper:check:unchanged	0.1233s	0.1226s	-0.57%	47.85 MiB	47.81 MiB	-0.09%
🟣 regex:check:unchanged	0.3025s	0.3009s	-0.54%	71.39 MiB	71.05 MiB	-0.49%
🟣 syn:check:unchanged	0.5760s	0.5734s	-0.44%	101.72 MiB	101.55 MiB	-0.17%
🟣 syntex_syntax:check:unchanged	1.4258s	1.4197s	-0.43%	202.54 MiB	202.23 MiB	-0.15%
Total	2.8392s	2.8262s	-0.46%	516.59 MiB	515.37 MiB	-0.24%
Summary	1.0000s	0.9951s	-0.49%	1 byte	1.00 bytes	-0.26%

[cjgillot]

@bors try @rust-timer queue

[bors]

⌛ Trying commit f8ad88b with merge df14cfe...

[bors]

☀️ Try build successful - checks-actions
Build commit: df14cfe (df14cfe1d6418344f82966df5d7589e1e8682dda)

[cjgillot]

@rust-timer build df14cfe

[rust-timer]

Finished benchmarking commit (df14cfe): comparison URL.

Overall result: no relevant changes - no 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.

@bors rollup=never
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

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

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)	3.3%	[2.1%, 5.4%]	3
Regressions ❌ (secondary)	2.8%	[2.8%, 2.8%]	1
Improvements ✅ (primary)	-3.1%	[-5.7%, -1.4%]	18
Improvements ✅ (secondary)	-3.3%	[-4.8%, -1.6%]	6
All ❌✅ (primary)	-2.2%	[-5.7%, 5.4%]	21

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)	2.6%	[0.8%, 3.7%]	6
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-4.1%	[-8.2%, -1.9%]	85
Improvements ✅ (secondary)	-7.4%	[-12.1%, -4.1%]	18
All ❌✅ (primary)	-3.6%	[-8.2%, 3.7%]	91

Binary size

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

Bootstrap: 631.075s -> 631.296s (0.04%)
Artifact size: 317.84 MiB -> 317.84 MiB (0.00%)

[cjgillot]

Good max-rss and cycles gains.
@bors r+

[bors]

📌 Commit f8ad88b has been approved by cjgillot

It is now in the queue for this repository.

[bors]

⌛ Testing commit f8ad88b with merge 3f5a62d...

[rust-log-analyzer]

The job armhf-gnu failed! Check out the build log: (web) (plain)

Click to see the possible cause of the failure (guessed by this bot)

failures:

---- [ui] tests/ui/const-generics/transparent-maybeunit-array-wrapper.rs#min stdout ----

error in revision `min`: test run failed!
status: exit status: 101
command: RUST_TEST_THREADS="8" "/checkout/obj/build/x86_64-unknown-linux-gnu/stage0-tools-bin/remote-test-client" "run" "0" "/checkout/obj/build/x86_64-unknown-linux-gnu/test/ui/const-generics/transparent-maybeunit-array-wrapper.min/a"
--- stdout -------------------------------
uploaded "/checkout/obj/build/x86_64-unknown-linux-gnu/test/ui/const-generics/transparent-maybeunit-array-wrapper.min/a", waiting for result
--- stderr -------------------------------
thread 'main' panicked at src/tools/remote-test-client/src/main.rs:310:9:
client.read_exact(&mut header) failed with Connection reset by peer (os error 104)
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

[bors]

💔 Test failed - checks-actions

[Zoxc]

That looks like a failure to retry.

[cjgillot]

@bors retry

[bors]

⌛ Testing commit f8ad88b with merge 341ef15...

[bors]

☀️ Test successful - checks-actions
Approved by: cjgillot
Pushing 341ef15 to master...

[rust-timer]

Finished benchmarking commit (341ef15): comparison URL.

Overall result: ❌ regressions - ACTION NEEDED

Next Steps: If you can justify the regressions found in this perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please open an issue or create a new PR that fixes the regressions, add a comment linking to the newly created issue or PR, and then add the perf-regression-triaged label to this PR.

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

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)	0.4%	[0.2%, 0.5%]	9
Regressions ❌ (secondary)	0.3%	[0.2%, 0.5%]	6
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	0.4%	[0.2%, 0.5%]	9

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)	4.3%	[2.4%, 6.2%]	2
Regressions ❌ (secondary)	2.5%	[1.5%, 3.5%]	3
Improvements ✅ (primary)	-2.7%	[-5.6%, -0.8%]	23
Improvements ✅ (secondary)	-3.5%	[-4.7%, -1.7%]	6
All ❌✅ (primary)	-2.1%	[-5.6%, 6.2%]	25

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)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-4.2%	[-8.1%, -1.4%]	90
Improvements ✅ (secondary)	-7.8%	[-13.3%, -4.4%]	18
All ❌✅ (primary)	-4.2%	[-8.1%, -1.4%]	90

Binary size

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

Bootstrap: 632.183s -> 632.945s (0.12%)
Artifact size: 318.14 MiB -> 318.26 MiB (0.04%)

[Kobzol]

The previous run didn't have any instruction count changes, but anyway the Max-RSS and especially cycle wins outweigh the few icount regressions.

@rustbot label: +perf-regression-triaged