One markdown file per benchmark run, one platform per file. See ../CONTRIBUTING.md for the naming convention and how to submit your own.
Each file is the literal output of the suite's Copy as markdown button — environment fingerprint plus per-test result tables (N, trimmed mean, median, p95, p99, stddev). No editing required.
Files are listed in NN order on the hosted index.