• jsonparser: https://github.com/buger/jsonparser
• jsoniter pull-api: https://github.com/json-iterator/go
• jsoniter reflect-api: https://github.com/json-iterator/go/blob/master/jsoniter_reflect.go
• encoding/json: golang standard lib
• easy json: https://github.com/mailru/easyjson

The goal is to prove jsoniter is not slow, not to prove it is the fastest, at least good enough. My motivation of inventing json iterator is the flexibility to mix high level and low level api. Performance is the by-product of schema based parsing.

• CPU: i7-6700K @ 4.0G
• Level 1 cache size: 4 x 32 KB 8-way set associative instruction caches
• Level 2 cache size: 4 x 256 KB 4-way set associative caches
• Level 3 cache size: 8 MB 16-way set associative shared cache
• Go: 1.8beta1

https://github.com/json-iterator/go-benchmark/blob/master/src/github.com/json-iterator/go-benchmark/benchmark_small_payload_test.go

pull-api is fast, and reflection-api is not slow either. encoding/json is not that slow on i7-6700K, but much slower on cpu with smaller cache.

https://github.com/json-iterator/go-benchmark/blob/master/src/github.com/json-iterator/go-benchmark/benchmark_medium_payload_test.go

reflect-api even out-performed the hand written parser

https://github.com/json-iterator/go-benchmark/blob/master/src/github.com/json-iterator/go-benchmark/benchmark_large_payload_test.go

This is a pure counting usage. jsonparser is faster

test file used: https://github.com/json-iterator/test-data/blob/master/large-file.json

The difference here is because jsonparser take []byte as input, but jsoniter can take io.Reader as input.