benchmark-chonk

To benchmark Chonk proving performance reliably, use pinned `ivc-inputs.msgpack` files captured from real end-to-end transaction flows. This ensures stable, comparable results by eliminating environmental variances, allowing you to accurately measure what changed and where execution time is spent.

Extract per-circuit operation timing by building the `bb` binary and executing `bb prove --scheme chonk` with `--print_bench` and `--bench_out_hierarchical` flags. For WASM runs, enable `BB_BENCH=1` to capture performance metrics across both environments.

Yes, Chonk proving profiling supports WASM environments via wasmtime with threads enabled. You can run benchmarks using `BB_BENCH=1` to measure WASM proving performance alongside native execution to ensure performance impacts are visible across both environments.

The best way to compare Chonk performance across code changes is using A/B comparisons against pinned IVC inputs. This method generates hierarchical benchmark breakdowns suitable for component-level analysis, keeping comparisons consistent over time.

You need pinned `ivc-inputs.msgpack` for performance regression testing because it captures real end-to-end transaction flows. Using pinned inputs keeps comparisons consistent over time, preventing external data variations from skewing your benchmark results.