Epic: Benchmark against Circom (Groth16) - Githubissues

noir-lang / zk_bench

Benchmark circom and noir on some standard primitives from circomlib/noir stdlib

MIT License

0 stars 0 forks source link

Epic: Benchmark against Circom (Groth16) #4

Closed Savio-Sou closed 1 month ago

Savio-Sou commented 3 months ago

Problem

We don't currently have comprehensive and objective insights into how Noir performs against other ZK tech stacks.

Happy case

### Standard Libraries
- [ ] Gather benchmark numbers on Nargo
- [ ] Gather benchmark numbers on NoirJS
- [ ] Gather benchmark numbers on snarkjs

### Semaphore
- [ ] Gather benchmark numbers on Nargo
- [ ] Gather benchmark numbers on NoirJS

Benchmarks should be conducted both for 1 and large number (e.g. 100) of iterations to put UltraPlonk's lookup table fixed cost into perspective.

References

Crypto team's benchmarking script: https://github.com/AztecProtocol/aztec-packages/blob/master/barretenberg/cpp/scripts/benchmark_wasm_remote.sh
Ze's NoirJS vs Circom Semaphore benchmarks: https://zpedro.notion.site/Semaphore-ad21b26922e442c6b988695717a108b1#e61659ceb7c04c09909c4902f2fbdac3
Savio's Nargo v0.21.0 benchmarks: https://github.com/Savio-Sou/noir-benchmarks
Community's Polylang vs Miden vs Risc0 vs Noir vs Leo benchmarks: https://zkbench.dev/