#Reproducible benchmark

    The benchmark invokes the real quote engine over the checked-in sample manifest 1,000 times. It is intentionally deterministic: there is no network, clock, or random input in the workload.

    Run it from the repository root:

    moon run benchmarks

    For wall-clock measurements, run the command three times with the same MoonBit toolchain and record the median elapsed time. The output checksum must remain:

    items=3000 total=CNY 269930.00

    Source Files