zp-arena-reset-free-all — arena-reset-free-all
Claim
ArenaAllocator.reset(.free_all) each iter vs .retain_capacity under Zig 0.17.0 -Ofast on this Linux machine shows a clear before/after wall-clock gap.
How measured
- Harness:
packages/zperf/examples/bench/evidence_fill.zig(rulezp-arena-reset-free-allcases) - Build:
zig build-exe -Ofast· optimize_mode=fast - Timing:
Io.Clockawake; warm=2; median of 9 rounds - before:
arena_reset_free_all - after:
arena_reset_retain - ops=50000
- Raw:
zp-arena-reset-free-all.jsonl
Runtime (this evidence file)
| Field | Value |
|---|---|
| When | 2026-10-21 Asia/Shanghai (JSONL ts is UTC) |
| Zig | 0.17.0 · fast |
| Median before | ~2.252 ms |
| Median after | ~0.7083 ms |
| ratio before/after | ~3.18× |
Lint mapping
zperf-lint zp-arena-reset-free-all / arena-reset-free-all. Default severity in rules.zig is warning; see evidence-policy for keep/escalate. Evidence exists and agrees directionally.
Verdict
Agree ~3.18× — free_all returns pages every time.
Limits
Use free_all when RSS must drop; retain for hot reuse.