Throughput Benchmarks

The current benchmark story for qail-zig is based on direct three-way comparisons between:

• pgx (Go)
• qail.rs (Rust)
• qail-zig

Harness

The active harness uses matching one-shot runners for:

• single — prepared single-query path on one connection
• pipeline — prepared batch pipeline on one connection
• pool10 — prepared singles over ten connections

Latest Isolated 12-Sample Medians

Reading the Results

• Zig currently leads on single and pool10 in this harness.
• Rust still leads on pipeline.
• pool10 is the noisiest mode and should always be read with variance, not just peak numbers.

Benchmark Discipline

The benchmark numbers are only meaningful when the compared paths are equivalent. The current work explicitly separated:

• interleaved round-robin runs
• isolated block runs
• per-round CSV/SVG graph output

That makes it easier to see whether a result comes from the implementation path or from order effects.