Closed chadbrewbaker closed 3 years ago
To benchmark the different implementations use:
cargo bench --bench=throughput_basic
cargo bench --bench=throughput_compat
cargo bench --bench=throughput_std
cargo bench --features=simdjson --bench=throughput_simdjson
Adding -- --save-baseline some_name
to the bench commandline and then using critcmp to compare benchmarks is handy as well.
I don't really use the cpb_
(cycles-per-byte) benchmarks at the moment. I will add some documentation and make cargo bench
do something useful or at least point to the docs.
Am I doing something stupid? Happy to add a PR to the docs on running the benchmark suite.
x86 WSL2 Ubuntu 20.04 environment using rust nightly toolchain