perf (tsort) : avoid reading the whole input into memory and intern strings

[anastygnome]

This PR is a WIP to get more performance out of tsort.

Move to usize is complete. Next step is to move to vec instead of hashmap

[codspeed-hq]

CodSpeed Performance Report

Merging #9872 will degrade performance by 5.29%

_{Comparing anastygnome:tsort (4420344) with main (84e6f03)}

Summary

⚡ 1 improvement
❌ 1 regression
✅ 128 untouched
⏩ 30 skipped¹

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	BASE	HEAD	Efficiency
⚡	tsort_input_parsing_heavy[5000]	84.2 ms	71.9 ms	+17.13%
❌	tsort_linear_chain[1000000]	1.5 s	1.6 s	-5.29%

Footnotes

1. 30 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[github-actions]

GNU testsuite comparison:

GNU test failed: tests/misc/tsort. tests/misc/tsort is passing on 'main'. Maybe you have to rebase?

[github-actions]

GNU testsuite comparison:

GNU test failed: tests/cp/cp-mv-enotsup-xattr. tests/cp/cp-mv-enotsup-xattr is passing on 'main'. Maybe you have to rebase?
Note: The gnu test tests/csplit/csplit-io-err was skipped on 'main' but is now failing.

[github-actions]

GNU testsuite comparison:

Congrats! The gnu test tests/tail/follow-name is no longer failing!

[github-actions]

GNU testsuite comparison:

Congrats! The gnu test tests/tail/follow-name is no longer failing!

[github-actions]

GNU testsuite comparison:

GNU test failed: tests/misc/tsort. tests/misc/tsort is passing on 'main'. Maybe you have to rebase?

[github-actions]

GNU testsuite comparison:

GNU test failed: tests/cp/cp-mv-enotsup-xattr. tests/cp/cp-mv-enotsup-xattr is passing on 'main'. Maybe you have to rebase?
Note: The gnu test tests/csplit/csplit-io-err was skipped on 'main' but is now failing.

[sylvestre]

any idea why the perf regressed here?

[anastygnome]

Oh, @sylvestre , I just added a few changes :)
The perf regression is most likely due to using Usize symbols. U32 actually improves performance.

[github-actions]

GNU testsuite comparison:

GNU test failed: tests/sort/sort-stale-thread-mem. tests/sort/sort-stale-thread-mem is passing on 'main'. Maybe you have to rebase?

[github-actions]

GNU testsuite comparison:

Congrats! The gnu test tests/tail/assert is no longer failing!

[anastygnome]

@sylvestre I believe the regression noticed are due to cache misses because the metadata that goes with the token is just slightly bigger than usize.

Here's another benchmark that maybe of interest:

graph.tsort, 1 Gb random graph 
Massif output: 
Gnu tsort: 

    MB (Memory)
248.6^                                                                       #
     |                                                          ::::::::::@::#
     |                                        ::::::::::::::::::: ::: ::::@: #
     |                               :::::@:::: ::::: : ::::: ::: ::: ::::@: #
     |                            @@:: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |                        ::::@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |                     @@::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |                   ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |               ::::::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |              :::: ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |            :::::: ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |         :::: :::: ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |       :::: : :::: ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |      :: :: : :::: ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |    :::: :: : :::: ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |    : :: :: : :::: ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |   :: :: :: : :::: ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     |   :: :: :: : :::: ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     | @@:: :: :: : :::: ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
     | @ :: :: :: : :::: ::@ ::: :@ :: : :@:: : ::::: : ::::: ::: ::: ::::@: #
   0 +----------------------------------------------------------------------->Gi (Giga Instructions) 
     0                                                                   23.47

Number of snapshots: 56
 Detailed snapshots: [1, 2, 16, 21, 26, 52, 55 (peak)]
 
 
 
uutils tsort 

     MB (Memory)
200.0^                                                                    #   
     |                                @::::@:::::::::::@::::@::::@::::@:::#:  
     |              :::::::::::@::::::@::::@::: :::::: @::::@::::@::::@:::#:  
     |           ::::::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:  
     |         :::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:: 
     |        ::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:: 
     |      @:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     |     :@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     |     :@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     |     :@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     |     :@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     |     :@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     |   @::@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     |  :@::@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     |  :@::@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     |  :@::@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     |  :@::@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     | ::@::@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     | ::@::@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
     | ::@::@:::::: :::::::::::@:::: :@::::@::: :::::: @::::@::::@::::@:::#:::
   0 +----------------------------------------------------------------------->Gi (Giga Instructions) 
     0                                                                   35.83

Number of snapshots: 98
 Detailed snapshots: [3, 6, 24, 31, 37, 50, 51, 52, 62, 72, 82, 91 (peak)]

Hyperfine : 
 
Benchmark 1: target/release/coreutils tsort /home/bench_vm/graph.tsort
  Time (mean ± σ):      6.044 s ±  0.071 s    [User: 5.602 s, System: 0.394 s]
  Range (min … max):    5.974 s …  6.168 s    10 runs
 
Benchmark 2: tsort /home/bench_vm/graph.tsort 
  Time (mean ± σ):      7.115 s ±  0.147 s    [User: 6.738 s, System: 0.328 s]
  Range (min … max):    6.910 s …  7.390 s    10 runs
`
Summary: 'target/release/coreutils tsort /home/bench_vm/graph.tsort'
ran 1.18 ± 0.03 times faster than 'tsort /home/bench_vm/graph.tsort'
 `

[github-actions]

GNU testsuite comparison:

Congrats! The gnu test tests/tail/assert is no longer failing!

[github-actions]

GNU testsuite comparison:

Congrats! The gnu test tests/tty/tty-eof is no longer failing!

[github-actions]

GNU testsuite comparison:

GNU test failed: tests/shuf/shuf-reservoir. tests/shuf/shuf-reservoir is passing on 'main'. Maybe you have to rebase?
GNU test failed: tests/sort/sort-stale-thread-mem. tests/sort/sort-stale-thread-mem is passing on 'main'. Maybe you have to rebase?
Skip an intermittent issue tests/timeout/timeout (fails in this run but passes in the 'main' branch)

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/timeout/timeout (fails in this run but passes in the 'main' branch)

[sylvestre]

windows fails with:

error[E0308]: mismatched types
  --> src\uu\tsort\src\tsort.rs:80:24
   |
80 |             return Err(TsortError::IsDir(input.to_string_lossy().to_string()));
   |                    --- ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ expected `UIoError`, found `TsortError`
   |                    |
   |                    arguments to this enum variant are incorrect
   |
help: the type constructed contains `TsortError` due to the type of the argument passed
  --> src\uu\tsort\src\tsort.rs:80:20
   |
80 |             return Err(TsortError::IsDir(input.to_string_lossy().to_string()));
   |                    ^^^^------------------------------------------------------^
   |                        |
   |                        this argument influences the type of `Err`
note: tuple variant defined here
  --> /rustc/8d670b93d40737e1b320fd892c6f169ffa35e49e/library\core\src\result.rs:566:4

[anastygnome]

@sylvestre fixed, the rest of the work will be put in another PR.

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/timeout/timeout (fails in this run but passes in the 'main' branch)

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/timeout/timeout (fails in this run but passes in the 'main' branch)

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/timeout/timeout (fails in this run but passes in the 'main' branch)

[anastygnome]

@sylvestre done :)

[anastygnome]

The performance regression is inevitable here if we want to support the same input range (2^64 -1 unique input tokens). There's also a slight buffering cost at the given bench graph size. We are still way more memory efficient, so I'd still consider it a net positive.