Pedersen Hash Benchmark

Comparison of performance of various Pedersen hash implementations.

Results

Results are divided into two categories: native and Node.js/WebAssembly. Node.js/WebAssembly benchmark results are only presented for Node.js as it has reasonably good WebAssembly performance with the V8 engine.

Native results

	pathfinder 👑	starknet-rs	starknet-signatures
pedersen_hash	27.858 µs	31.990 µs	143.21 µs
Relative	1.00x	1.15x	5.14x

Node.js/WebAssembly results

	starknet-rs 👑	micro-starknet
pedersen_hash	228.67 µs	1.573 ms
Relative	1.00x	6.88x

Implementations

Here's the list of implementations tested along with their platform availability.

Implementation	Native	Node.js/WebAssembly
geometryresearch/starknet-signatures	✅	❌
eqlabs/pathfinder	✅	❌
xJonathanLEI/starknet-rs	✅	✅
paulmillr/micro-starknet	❌	✅

Environment

• CPU

  AMD Ryzen 9 5950X 16-Core Processor

• OS

  Ubuntu 20.04.5 LTS

• Runtimes

  • wasmer-js: wasmer-js 0.4.1
  • Node.js: v18.12.1

Running benchmarks

Native benchmarks

To run the native benchmarks:

$ cargo bench

WebAssembly benchmarks

To run the WebAssembly benchmarks, make sure you have wasm32-wasi target and cargo-wasi installed. Then build the wasm file:

$ ./scripts/build_bench_wasm.sh

Then you can run the benchmark with your target runtime. For example, to run the benchmark against wasmer-js (Node.js):

$ ./scripts/run_bench_wasm.sh wasmer-js

For more information regarding wasm benchmarks with criterion.rs, check out this guide.

Node.js benchmarks

To run non-wasm Node.js benchmarks, first build the benchmark script:

$ yarn install
$ yarn build

Then run the script with node:

$ node build/js-bench.js