Other languages
Weekly measurement runs ↗| Implementation | Measured value | How Native compares |
|---|
Measurement details and scope
This small-input comparison updates on Sunday around 05:23 JST when performance inputs changed, or on demand. It is independent of daily measurements and PR acceptance. Each weekly run includes current Native on the same host; values from different runs are never divided to produce comparisons.
The five external implementations are the same registered upstream sources used by the detailed comparison. Program structure and optimization choices contribute to the results. All six implementations share one logical CPU and checked output oracles. One warmup and five retained fresh processes include startup; Java JIT and JVM initialization occur in every process. This is not Java steady-state throughput or a general language ranking. Peak RSS is GNU time maximum RSS, not a process-tree sum.
Compiler warnings remain in raw build evidence. Runtime output, stderr, exits and timeouts are checked. Failed observations stay visible. Toolchain versions, source and executable hashes, inputs and raw samples accompany each run. Measurement strategy ↗