On Tue, Apr 02, 2019 at 08:20:05PM +0200, Johannes Schindelin wrote: > > Hitting some quadratic complexity algorithm explains why the "degradation" > > is so huge. And if I understand it correctly we are more or less "unlucky" > > to hit such access patterns more often in the new repo. > > > > Anyway, nice to see that you already have some ideas regarding how to > > improve this. > > In the meantime, maybe the commit graph feature can help you? > > git config --global core.commitGraph true > git config --global gc.writeCommitGraph true > git commit-graph write --reachable > > The idea being that the commit graph extracts a lot of information > pre-emptively and makes the lookup of those bits faster than unpacking the > commits all the time. I'd definitely encourage Björn (or anybody with a biggish repo) to check out the commit-graph work, because it's really cool. But it won't help much here, because the quadratic loop is just on linked-list insertion. We're literally spending all of our time chasing pointers in memory. -Peff
