- To understand the full benchmark flow, read Benchmark Execution.
- To understand the canonical EEST JSON accepted by
ere-hosts, read Benchmark Execution Inputs. - To understand the files and JSON written by
ere-hosts, read Benchmark Execution Output. - To publish canonical stateless input batches as a public R2 dataset or validate that dataset locally with EEST, read Stateless Input Publication.
- Benchmark Execution: common
ere-hostscommands, action-aware input requirements, proof verification, guest artifacts, and operational notes. - Benchmark Execution Inputs: canonical EEST schema, input discovery, fixture filtering, execution-client routing, and legacy-format rejection.
- Benchmark Execution Output: metrics directory layout,
BenchmarkRunJSON, hardware metadata, proof handling, and input dumps. - Stateless Input Publication: the separate
witness-generator-spec-cliR2 publication flow, download examples, systemd assets, and local EEST validation.
The zkevm-metrics crate API documentation lives in crates/metrics/README.md. The CLI metrics files written by ere-hosts are documented in Benchmark Execution Output.