Remove extra lookups and memory allocations from tsort graph construction

[Nekrolm]

Hi.
I was benchmarking several tools and found there some low hanging minor optimizations for input processing that can be done:

• Don't lookup in hashmap twice
• Don't allocate extra arrays of input strings

[sylvestre]

did you look at the actual perf wins ?
(with hyperfine - previous rust impl, gnu and this one)

[github-actions]

GNU testsuite comparison:

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

[sylvestre]

a job is failing, you need to do
cd fuzz && cargo build
to update Cargo.lock (yes, it is silly)

[Nekrolm]

It optimizes only initial input processing, so any gain can be visible only on large files. Where the initial read is somehow noticeable

With 1 million edges linear chain (edges shuffled)

Benchmark 1: target/release/coreutils tsort line_shuf.txt || true
  Time (mean ± σ):     19.849 s ±  1.410 s    [User: 17.447 s, System: 2.365 s]
  Range (min … max):   17.963 s … 21.943 s    10 runs

After

Benchmark 1: target/release/coreutils tsort line_shuf.txt || true
  Time (mean ± σ):     19.577 s ±  0.652 s    [User: 17.293 s, System: 2.262 s]
  Range (min … max):   18.920 s … 20.719 s    10 runs

Actually, Itertools are not needed. And it's even better without them -- internal chunks' buffer overuses RefCell runtime checks
I've updated PR without it:

No itertools

Benchmark 1: target/release/coreutils tsort line_shuf.txt || true
  Time (mean ± σ):     18.866 s ±  0.453 s    [User: 16.725 s, System: 2.137 s]
  Range (min … max):   18.289 s … 19.867 s    10 runs
 

tsort (GNU coreutils) 9.4 -- (have't checked latest one)

dmis@dmis-asus-N7600PC:~/WORKSPACE/coreutils$ hyperfine "tsort line_shuf.txt || true"
Benchmark 1: tsort line_shuf.txt || true
  Time (mean ± σ):     28.474 s ±  4.092 s    [User: 27.882 s, System: 0.498 s]
  Range (min … max):   25.950 s … 39.779 s    10 runs

[sylvestre]

please run hyperfine this way:
hyperfine "tsort line_shuf.txt || true" " target/release/coreutils tsort line_shuf.txt || true"
it work better this way to compare implementations

[codspeed-hq]

CodSpeed Performance Report

Merging #8694 will not alter performance

_{Comparing Nekrolm:main (4f09383) with main (7dbeb8f)}

Summary

✅ 55 untouched
⏩ 1 skipped¹

Footnotes

1. 1 benchmark was skipped, so the baseline result was used instead. If it was deleted from the codebase, click here and archive it to remove it from the performance reports. ↩

[Nekrolm]

dmis@dmis-asus-N7600PC:~/WORKSPACE/coreutils$  hyperfine "tsort line_shuf.txt || true" " target/release/coreutils tsort line_shuf.txt || true"
Benchmark 1: tsort line_shuf.txt || true
  Time (mean ± σ):     33.722 s ±  3.816 s    [User: 33.083 s, System: 0.620 s]
  Range (min … max):   27.813 s … 40.324 s    10 runs
 
Benchmark 2:  target/release/coreutils tsort line_shuf.txt || true
  Time (mean ± σ):     21.824 s ±  2.108 s    [User: 19.456 s, System: 2.356 s]
  Range (min … max):   18.467 s … 25.517 s    10 runs
 
Summary
   target/release/coreutils tsort line_shuf.txt || true ran
    1.55 ± 0.23 times faster than tsort line_shuf.txt || true

dmis@dmis-asus-N7600PC:~/WORKSPACE/coreutils$ cat /proc/cpuinfo 
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 140
model name      : 11th Gen Intel(R) Core(TM) i7-11370H @ 3.30GHz
stepping        : 1
microcode       : 0xbc
cpu MHz         : 843.122
cache size      : 12288 KB
physical id     : 0
siblings        : 8
core id         : 0
cpu cores       : 4
apicid          : 0
initial apicid  : 0
fpu             : yes
fpu_exception   : yes
cpuid level     : 27
wp              : yes

[sylvestre]

well done!

[github-actions]

GNU testsuite comparison:

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

[sylvestre]

maybe document this a bit more to explain what it is doing ;)

[Nekrolm]

Expanded a bit

[github-actions]

GNU testsuite comparison:

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

[github-actions]

GNU testsuite comparison:

Skipping an intermittent issue tests/misc/tee (passes in this run but fails in the 'main' branch)

[github-actions]

GNU testsuite comparison:

GNU test failed: tests/tail/overlay-headers. tests/tail/overlay-headers is passing on 'main'. Maybe you have to rebase?
Skipping an intermittent issue tests/timeout/timeout (passes in this run but fails in the 'main' branch)

[github-actions]

GNU testsuite comparison:

GNU test failed: tests/tail/overlay-headers. tests/tail/overlay-headers is passing on 'main'. Maybe you have to rebase?

[github-actions]

GNU testsuite comparison:

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