高性能的序列化和反序列化算法库xxxPack（未想好名称）
对比目前业界高性能库：
- encoding/gob
- encoding/json
- github.com/vmihailenco/msgpack/v5
- labix.org/v2/mgo/bson
- github.com/valyala/fastjson
- github.com/json-iterator/go
- 自研序列化算法
验证数据集合：
type Data struct {
Name string
BirthDay time.Time
Phone string
Siblings int
Spouse bool
Money float64
ExInfo interface{}
}
Benchmark Results
2023-02-09 Results with
Go 1.19.4MacPro Apple M2 8G macOS VenturaDarwin 192.168.1.7 22.2.0 Darwin Kernel Version 22.2.0: Fri Nov 11 02:06:26 PST 2022; root:xnu-8792.61.2~4/RELEASE_ARM64_T8112 arm64
环境下,数据：
goos: darwin
goarch: amd64
pkg: github.com/kubeservice-stack/serialization-benchmarks
cpu: VirtualApple @ 2.50GHz
Benchmark_Bson_Marshal-8 2272702 499.2 ns/op 110.0 B/serial 376 B/op 10 allocs/op
Benchmark_Bson_Unmarshal-8 1617357 716.2 ns/op 110.0 B/serial 224 B/op 19 allocs/op
Benchmark_FastJson_Marshal-8 4016576 296.9 ns/op 133.8 B/serial 504 B/op 6 allocs/op
Benchmark_FastJson_Unmarshal-8 1527366 778.7 ns/op 133.8 B/serial 1704 B/op 9 allocs/op
Benchmark_Gob_Marshal-8 467408 2475 ns/op 163.6 B/serial 1616 B/op 35 allocs/op
Benchmark_Gob_Unmarshal-8 95322 12386 ns/op 163.6 B/serial 7688 B/op 207 allocs/op
Benchmark_JsonIter_Marshal-8 1941692 591.5 ns/op 148.7 B/serial 216 B/op 3 allocs/op
Benchmark_JsonIter_Unmarshal-8 1307601 871.8 ns/op 148.7 B/serial 359 B/op 14 allocs/op
Benchmark_Json_Marshal-8 1406618 850.8 ns/op 148.6 B/serial 208 B/op 2 allocs/op
Benchmark_Json_Unmarshal-8 494430 2153 ns/op 148.6 B/serial 399 B/op 9 allocs/op
Benchmark_xxxPack_Marshal-8 3624224 312.8 ns/op 119.0 B/serial 344 B/op 6 allocs/op
Benchmark_xxxPack_Unmarshal-8 2070897 574.7 ns/op 119.0 B/serial 112 B/op 3 allocs/op
Benchmark_Msgpack_Marshal-8 2961548 401.9 ns/op 92.00 B/serial 264 B/op 4 allocs/op
Benchmark_Msgpack_Unmarshal-8 2209035 536.9 ns/op 92.00 B/serial 160 B/op 4 allocs/op
PASS
ok github.com/kubeservice-stack/serialization-benchmarks 23.928s
总结
- xxxPack
一次序列化+一次反序列化，性能最优综合排第1 ：574.7 ns/op + 312.8 ns/op = 887.5 ns/op - xxxPack 整体性能是 golang原生json库的 3.38倍* ：（2153+850.8）/ 887.5 = 3.3845倍
- xxxPack 每次申请操作平均内存申请大小（119B/serial）排 第3； xxxPack Unmarshal 内存申请最小（112 B/op）排第1
- xxxPack Unmarshal 每次操作申请了内存次数 排第2； Marshal 排第4;
「如果这篇文章对你有用,请随意打赏」
如果这篇文章对你有用,请随意打赏
使用微信扫描二维码完成支付
