FWIW, I see this problem too. In my case, I'm working with a repo that was converted from Subversion. It has some rather large binary data in it too (~400mb file that has a custom toolchain). Several versions of it were checked in over time. I'd expect that we shouldn't need to do much server-side. I mean we need the graph, but we shouldn't revisions be transferred in a more streamy fashion?
FWIW, I see this problem too. In my case, I'm working with a repo that was converted from Subversion. It has some rather large binary data in it too (~400mb file that has a custom toolchain). Several versions of it were checked in over time. I'd expect that we shouldn't need to do much server-side. I mean we need the graph, but we shouldn't revisions be transferred in a more streamy fashion?