This is a partial port of nativejson-benchmark to Rust. The libraries tested are:
- serde_json 1.0.72
- json-rust 0.12.4
- rustc-serialize 0.3.24
- simd-json 0.4.11 (this requires a modern x86 CPU for good results)
DOM STRUCT
======= serde_json ======= parse|stringify ===== parse|stringify ====
data/canada.json 320 MB/s 430 MB/s 580 MB/s 310 MB/s
data/citm_catalog.json 420 MB/s 560 MB/s 710 MB/s 880 MB/s
data/twitter.json 300 MB/s 910 MB/s 550 MB/s 1060 MB/s
======= json-rust ======== parse|stringify ===== parse|stringify ====
data/canada.json 390 MB/s 840 MB/s
data/citm_catalog.json 520 MB/s 780 MB/s
data/twitter.json 430 MB/s 1030 MB/s
==== rustc_serialize ===== parse|stringify ===== parse|stringify ====
data/canada.json 150 MB/s 67 MB/s 120 MB/s 46 MB/s
data/citm_catalog.json 210 MB/s 180 MB/s 140 MB/s 210 MB/s
data/twitter.json 120 MB/s 330 MB/s 87 MB/s 350 MB/s
======= simd-json ======== parse|stringify ===== parse|stringify ====
data/canada.json 380 MB/s 470 MB/s 580 MB/s
data/citm_catalog.json 720 MB/s 760 MB/s 1220 MB/s
data/twitter.json 810 MB/s 890 MB/s 1050 MB/s
- Intel(R) Core(TM) i7-6600U CPU @ 2.60GHz (laptop CPU from 2015)
- rustc 1.59.0-nightly (0b6f079e4 2021-12-07)
To update the numbers above, I run ./json-benchmark
twice on an otherwise idle
computer and take the greater of the two results for each number.
For comparison, here are results from RapidJSON on the same hardware with the
nativejson-benchmark modified to run 256 times instead of 10. Code is in the
cpp
directory of this repo.
==== rapidjson-clang ===================== parse|stringify ====
data/canada.json 380 MB/s 200 MB/s
data/citm_catalog.json 680 MB/s 1090 MB/s
data/twitter.json 320 MB/s 600 MB/s
===== rapidjson-gcc ====================== parse|stringify ====
data/canada.json 390 MB/s 250 MB/s
data/citm_catalog.json 890 MB/s 1650 MB/s
data/twitter.json 440 MB/s 870 MB/s
- clang version 11.0.0-++20200128080810+b96e6859c99-1
exp120200128191406.2758 - gcc version 10.1.0 (Ubuntu 10.1.0-2ubuntu1~18.04)
Licensed under either of Apache License, Version 2.0 or MIT license at your option.
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in this project by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.