Test out a more lock-free hash map

[Zoxc]

This tests out the more lock-free hash map from my horde crate by using it as the default query map. The integration with DynSync / DynSend in this PR is unsound.

Shards are no longer used for the map, which means all writes are behind a single lock. Reads are lock-free.

[rustbot]

r? @Nadrieril

rustbot has assigned @Nadrieril.
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

[lqd]

@bors try @rust-timer queue

[bors]

⌛ Trying commit 2947bae with merge b73204a...

[bors]

☔ The latest upstream changes (presumably #138414) made this pull request unmergeable. Please resolve the merge conflicts.

[bors]

☀️ Try build successful - checks-actions
Build commit: b73204a (b73204a7a915c62fc1367f8bdf8f3ff971d2e4c6)

[rust-timer]

Finished benchmarking commit (b73204a): 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)	0.2%	[0.2%, 0.3%]	2
Regressions ❌ (secondary)	1.8%	[1.8%, 1.9%]	6
Improvements ✅ (primary)	-0.4%	[-1.3%, -0.1%]	35
Improvements ✅ (secondary)	-0.9%	[-1.6%, -0.2%]	35
All ❌✅ (primary)	-0.4%	[-1.3%, 0.3%]	37

Max RSS (memory usage)

Results (secondary 5.1%)

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)	6.7%	[3.0%, 8.0%]	4
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-1.3%	[-1.3%, -1.3%]	1
All ❌✅ (primary)	-	-	0

Cycles

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

Binary size

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

Bootstrap: 779.458s -> 783.14s (0.47%)
Artifact size: 365.27 MiB -> 362.93 MiB (-0.64%)

[Zoxc]

Local test with 8 threads / 8 cores:

Benchmark	Before	After		Before	After		Before	After
	Time	Time	%	Physical Memory	Physical Memory	%	Committed Memory	Committed Memory	%
🟣 clap:check	0.4593s	0.4569s	-0.52%	203.55 MiB	202.20 MiB	-0.67%	270.26 MiB	268.65 MiB	-0.60%
🟣 hyper:check	0.1369s	0.1360s	-0.61%	128.02 MiB	127.29 MiB	-0.57%	192.46 MiB	191.54 MiB	-0.47%
🟣 regex:check	0.2915s	0.2898s	-0.59%	163.22 MiB	162.03 MiB	-0.73%	222.23 MiB	220.81 MiB	-0.64%
🟣 syn:check	0.5602s	0.5591s	-0.19%	198.26 MiB	197.50 MiB	-0.38%	258.15 MiB	257.16 MiB	-0.38%
Total	1.4478s	1.4418s	-0.42%	693.05 MiB	689.01 MiB	-0.58%	943.09 MiB	938.16 MiB	-0.52%
Summary	1.0000s	0.9952s	-0.48%	1 byte	0.99 bytes	-0.59%	1 byte	0.99 bytes	-0.52%

Local test with 16 threads / 8 cores:

Benchmark	Before	After		Before	After		Before	After
	Time	Time	%	Physical Memory	Physical Memory	%	Committed Memory	Committed Memory	%
🟣 clap:check	0.4308s	0.4243s	💚 -1.51%	210.82 MiB	209.18 MiB	-0.78%	278.63 MiB	276.88 MiB	-0.63%
🟣 hyper:check	0.1350s	0.1336s	💚 -1.03%	131.61 MiB	130.79 MiB	-0.63%	196.52 MiB	195.54 MiB	-0.50%
🟣 regex:check	0.2834s	0.2795s	💚 -1.37%	168.81 MiB	167.36 MiB	-0.86%	228.35 MiB	226.61 MiB	-0.76%
🟣 syn:check	0.5418s	0.5379s	-0.71%	204.90 MiB	204.09 MiB	-0.39%	264.86 MiB	263.87 MiB	-0.38%
Total	1.3910s	1.3754s	💚 -1.13%	716.14 MiB	711.42 MiB	-0.66%	968.35 MiB	962.89 MiB	-0.56%
Summary	1.0000s	0.9884s	💚 -1.16%	1 byte	0.99 bytes	-0.66%	1 byte	0.99 bytes	-0.57%

It does seem to reduce congestion even with the single write lock.

[Zoxc]

I added a commit that uses this for type interning also.

Local test with 16 threads / 8 cores for just the type interning change:

Benchmark	Before	After		Before	After		Before	After
	Time	Time	%	Physical Memory	Physical Memory	%	Committed Memory	Committed Memory	%
🟣 clap:check	0.4460s	0.4432s	-0.62%	208.13 MiB	165.73 MiB	💚 -20.37%	276.49 MiB	215.77 MiB	💚 -21.96%
🟣 hyper:check	0.1399s	0.1293s	💚 -7.53%	129.73 MiB	89.35 MiB	💚 -31.12%	195.33 MiB	131.10 MiB	💚 -32.89%
🟣 regex:check	0.2890s	0.2789s	💚 -3.47%	166.27 MiB	122.74 MiB	💚 -26.18%	226.41 MiB	163.21 MiB	💚 -27.92%
🟣 syn:check	0.5584s	0.5499s	💚 -1.53%	203.10 MiB	159.39 MiB	💚 -21.53%	263.71 MiB	200.96 MiB	💚 -23.79%
Total	1.4333s	1.4014s	💚 -2.22%	707.23 MiB	537.21 MiB	💚 -24.04%	961.94 MiB	711.03 MiB	💚 -26.08%
Summary	1.0000s	0.9671s	💚 -3.29%	1 byte	0.75 bytes	💚 -24.80%	1 byte	0.73 bytes	💚 -26.64%

It does seem like sharding wasted quite a bit of memory.

[lqd]

We can do another perf run if you’d like, but the conflicts would need to be resolved.