Skip to content

zstd performance and Reader/Writer reuse #248

@vmihailenco

Description

@vmihailenco

I have a benchmark that produces following results:

BenchmarkCZstd-12      	     664	   1793805 ns/op	 125.62 MB/s	  522419 B/op	       9 allocs/op
BenchmarkGoZstd-12    	     374	   3247447 ns/op	  69.39 MB/s	 1646825 B/op	      12 allocs/op

That is roughly 2 times slower (and results can be worse on smaller payloads) and allocs are rather big (1646825 to encode & decode 225339 bytes?). I know that I can achieve better results by using EncodeAll and DecodeAll, but I would like to use Encoder/Decoder as wrappers over bytes.Buffer. So my question is - am I doing anything wrong here with Encoder/Decoder? I've disabled CRC for more fair comparison based on our previous discussion - I plan to use it in production.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions