This PR bootstraps a simple benchmarking infrastructure for comparing the performance of three different GFA analysis implementations (odgi, slow-odgi, and fgfa). Not that these results are all that interesting for the moment, but hopefully some of this stuff makes it possible to measure real differences once they exist?
There's a README that describes how to use these things. It can produce hilarious bar graphs like this:
That's comparing odgi paths with fgfa paths on a few GFAs (pre-converted to each tool's native format). In this run, odgi paths takes 3–12 seconds while fgfa paths takes 3–6 milliseconds (because it has no serialization step, and it never even reads the entire file).
This PR bootstraps a simple benchmarking infrastructure for comparing the performance of three different GFA analysis implementations (odgi, slow-odgi, and fgfa). Not that these results are all that interesting for the moment, but hopefully some of this stuff makes it possible to measure real differences once they exist?
There's a README that describes how to use these things. It can produce hilarious bar graphs like this:
That's comparing
odgi pathswithfgfa pathson a few GFAs (pre-converted to each tool's native format). In this run,odgi pathstakes 3–12 seconds whilefgfa pathstakes 3–6 milliseconds (because it has no serialization step, and it never even reads the entire file).