PuffAligner significantly slower at mapping and thread usage is much less than specified when running with `--compressedOutput`
Opened this issue · 1 comments
hermidalc commented
I've test both with --compressedOutput and without, and surprised how much slower align is when it is activated, writing of the compressed SAM file becomes a bottleneck to the entire command, and thread usage is much lower than what is specified. This is surprising as I haven't see writing to a compressed stream to cause such a bottleneck and change in performance with other tools.
hermidalc commented
It's much faster to not use --compressedOutput and pigz the SAM file afterwards. Then align uses all the threads that are specified and maps at a much faster speed