Native Benchmarks
This native benchmark (opens in a new tab) is based on the Rust implementation of each crate.
- Conducted on a 2020 M1 MacBook Pro 13-inch, dated 2024-05-06.
- The tasks with names starting with
automerge
use the automerge paper dataset. - In this benchmark, compression is disabled for both automerge and loro.
- Diamond-type doesn't support the list type yet.
Tasks | automerge | loro | diamond-type | yrs |
---|---|---|---|---|
automerge - apply | 557.42 ± 0.02 ms | 76.31 ± 0.03 ms | 17.68 ± 0.04 ms | 4239.71 ± 65.33 ms |
automerge - decode time | 767.01 ± 3.40 ms | 4.44 ± 0.05 ms | 2.53 ± 0.01 ms | 3.97 ± 0.01 ms |
automerge - encode time | 11.43 ± 0.004 ms | 4.58 ± 0.10 ms | 1.26 ± 0.004 ms | 857.76 ± 96.03 us |
concurrent list inserts | 77.15 ± 0.35 ms | 127.11 ± 0.79 ms | 64.35 ± 0.22 ms | 13.12 ± 0.11 ms |
list_random_insert_1k | 256.27 ± 0.97 ms | 11.48 ± 0.37 ms | 3.68 ± 0.004 ms | 4.48 ± 0.005 ms |