Golang's net/textproto but 𝙁/𝘼/𝙎/𝙏/𝙀/𝙍

import "gopkg.in/textproto.v0"

The Golang standard library net/textproto suffers in data throughput performance when using its DotReader or DotWriter implementations. It's caused by iterating each input output byte using the ReadByte or WriteByte functions. This increases a lot of CPU cycles for accessing memory for each byte. This module aim to provide an optimized implementation by leveraging the SIMD optimized ReadSlice and IndexByte functions. Here are the resulting numbers:

BenchmarkDotReader/Legacy-10                  72          15078444 ns/op         278.17 MB/s
BenchmarkDotReader/Optimized-10             1752            653339 ns/op        6419.79 MB/s
BenchmarkDotWriter/Legacy-10                 100          10760824 ns/op         389.78 MB/s
BenchmarkDotWriter/Optimized-10             2116            553493 ns/op        7577.88 MB/s