Cat splice perf

[FidelSch]

Based on #10832

Avoids using splice on small files.
May result in a small perf improvement

Threshold set to 16KB, determined by trial and error:

 hyperfine -w 10 -i -L coreutils "target/release/coreutils_16kb","target/release/coreutils_32kb","target/release/coreutils_128kb","target/release/coreutils_base"  "{coreutils} cat /tmp/threshold_test/file_16K*" -N
Benchmark 1: target/release/coreutils_16kb cat /tmp/threshold_test/file_16K*
  Time (mean ± σ):       1.4 ms ±   0.5 ms    [User: 0.3 ms, System: 1.0 ms]
  Range (min … max):     0.8 ms …  11.2 ms    2089 runs
 
  Warning: Ignoring non-zero exit code.
  Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
 
Benchmark 2: target/release/coreutils_32kb cat /tmp/threshold_test/file_16K*
  Time (mean ± σ):       1.5 ms ±   0.4 ms    [User: 0.3 ms, System: 1.1 ms]
  Range (min … max):     0.8 ms …   3.3 ms    3211 runs
 
  Warning: Ignoring non-zero exit code.
 
Benchmark 3: target/release/coreutils_128kb cat /tmp/threshold_test/file_16K*
  Time (mean ± σ):       1.4 ms ±   0.4 ms    [User: 0.3 ms, System: 1.0 ms]
  Range (min … max):     0.7 ms …   4.1 ms    2657 runs
 
  Warning: Ignoring non-zero exit code.
 
Benchmark 4: target/release/coreutils_base cat /tmp/threshold_test/file_16K*
  Time (mean ± σ):       1.6 ms ±   0.5 ms    [User: 0.3 ms, System: 1.2 ms]
  Range (min … max):     0.7 ms …   5.1 ms    2113 runs
 
  Warning: Ignoring non-zero exit code.
 
Summary
  target/release/coreutils_16kb cat /tmp/threshold_test/file_16K* ran
    1.02 ± 0.48 times faster than target/release/coreutils_128kb cat /tmp/threshold_test/file_16K*
    1.14 ± 0.50 times faster than target/release/coreutils_32kb cat /tmp/threshold_test/file_16K*
    1.17 ± 0.53 times faster than target/release/coreutils_base cat /tmp/threshold_test/file_16K*

[FidelSch]

Still slower than GNU

[oech3]

How about splice threshould + clap bypass?
Also please fix clippy.

[github-actions]

GNU testsuite comparison:

Congrats! The gnu test tests/tail/tail-n0f is now passing!

[sylvestre]

please run the benchmark with /usr/bin/cat too

[FidelSch]

now running hyperfine using --show-output, as otherwise perf improvement does not make much sense when it comes to writes (every command takes about the same). Downside being it seems to introduce significant noise:

hyperfine -w 10  -L coreutils "target/release/cat_16k","target/release/cat_64k","target/release/cat_bypass","target/release/cat_base","target/release/cat_bypass_8k","/usr/bin/cat"  "{coreutils} /tmp/threshold_test/file_16K*" --show-output
Benchmark 1: target/release/cat_16k /tmp/threshold_test/file_16K*
  Time (mean ± σ):     347.5 ms ± 209.5 ms    [User: 1.3 ms, System: 2.0 ms]
  Range (min … max):   123.6 ms … 713.3 ms    10 runs
 
Benchmark 2: target/release/cat_64k /tmp/threshold_test/file_16K*
  Time (mean ± σ):     125.3 ms ± 134.1 ms    [User: 0.3 ms, System: 1.8 ms]
  Range (min … max):    68.3 ms … 704.3 ms    21 runs
 
  Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
 
Benchmark 3: target/release/cat_bypass /tmp/threshold_test/file_16K*
  Time (mean ± σ):      77.5 ms ±  16.7 ms    [User: 0.4 ms, System: 0.9 ms]
  Range (min … max):    44.6 ms … 101.3 ms    33 runs
 
Benchmark 4: target/release/cat_base /tmp/threshold_test/file_16K*
  Time (mean ± σ):     109.1 ms ± 221.0 ms    [User: 0.3 ms, System: 1.1 ms]
  Range (min … max):    44.7 ms … 1710.4 ms    55 runs
 
  Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
 
Benchmark 5: target/release/cat_bypass_8k /tmp/threshold_test/file_16K*
  Time (mean ± σ):      82.8 ms ±  23.6 ms    [User: 0.3 ms, System: 1.3 ms]
  Range (min … max):    52.9 ms … 154.0 ms    18 runs
 
Benchmark 6: /usr/bin/cat /tmp/threshold_test/file_16K*
  Time (mean ± σ):     142.9 ms ± 212.9 ms    [User: 0.3 ms, System: 1.2 ms]
  Range (min … max):    54.0 ms … 1130.2 ms    27 runs
 
  Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
 
Summary
  target/release/cat_bypass /tmp/threshold_test/file_16K* ran
    1.07 ± 0.38 times faster than target/release/cat_bypass_8k /tmp/threshold_test/file_16K*
    1.41 ± 2.87 times faster than target/release/cat_base /tmp/threshold_test/file_16K*
    1.62 ± 1.77 times faster than target/release/cat_64k /tmp/threshold_test/file_16K*
    1.84 ± 2.78 times faster than /usr/bin/cat /tmp/threshold_test/file_16K*
    4.48 ± 2.87 times faster than target/release/cat_16k /tmp/threshold_test/file_16K*

[oech3]

Does this fix #9609 too?

[github-actions]

GNU testsuite comparison:

GNU test failed: tests/pr/bounded-memory. tests/pr/bounded-memory is passing on 'main'. Maybe you have to rebase?
Note: The gnu test tests/expand/bounded-memory is now being skipped but was previously passing.

[codspeed-hq]

Merging this PR will not alter performance

✅ 284 untouched benchmarks
⏩ 38 skipped benchmarks¹

---

_{Comparing FidelSch:cat-splice-perf (4102ac3) with main (ec7e81e)}

Footnotes

1. 38 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. ↩

[FidelSch]

Does this fix #9609 too?

As I understand, this should fix the specific example presented, but I frankly don't know if a similar case could arrive in the future. All it takes is for stat() to wrongly report some size larger that the defined threshold.

[github-actions]

GNU testsuite comparison:

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