Closed sampsyo closed 6 months ago
The next bottleneck in GFA parsing was the external rs-gfa library. I replaced this with a hand-rolled one.
Using the same measurement setup as #153:
So that's another 1.9x and 1.5x speedup over the last set of optimizations, for a total of 4x speedup over the first version.
It's clear that the bottleneck now is in the memcpying to the destination files, which is also avoidable (with some compromises).
memcpy
The next bottleneck in GFA parsing was the external rs-gfa library. I replaced this with a hand-rolled one.
Using the same measurement setup as #153:
So that's another 1.9x and 1.5x speedup over the last set of optimizations, for a total of 4x speedup over the first version.
It's clear that the bottleneck now is in the
memcpy
ing to the destination files, which is also avoidable (with some compromises).