←Older revision | Newer revision→
Our compression implementation (LZ, GLC, QUIC) are not optimized to current processors. There are speed gains to be had. Also revealed as a bottleneck while profiling.
- Hans de Goede
- Targeted Release: Not set
- 0% complete
There are existing drop in oss libraries that implement this, as a first step try to use them and see if we get a speed bonus.
TODO - add specific tests and results here, as a baseline.