bench

name	bench
description	Run the SortingBenchmark and report the vEB-sort vs qsort timing comparison. Use when the user wants a perf number after changes to vebtrees.h.

/bench

Runs the SortingBenchmark test binary, which sorts 500k dense keys (averaged over 100 runs) using both a vEB tree and stdlib qsort, and prints both timings.

Steps

1. Check whether build/test/SortingBenchmark exists. If not, run ./build.sh first (or suggest /build).
2. Run build/test/SortingBenchmark.
3. Parse the two "took X milliseconds" lines.

Report

• Both timings, plus the ratio (vEB vs qsort).
• The expected ballpark is ~6 ms vs ~60 ms — roughly a 10× speedup per the README.
• Flag loudly if vEB is slower than qsort, or if the ratio has regressed meaningfully vs what the README documents. That's a perf regression and worth investigating before the change lands.
• One-shot numbers are fine for iteration. If the user wants a more stable number, rerun a few times — clock noise on a ~6 ms measurement is non-trivial.

Notes

• The benchmark also asserts that the sorted output is strictly increasing. If an assert fires, the sorting logic is broken — not a perf issue.
• Benchmark binary ignores any CLI args today.