I'm just now pushing a patch that reduces the push memory use from 160M down to 45M. It's almost irrelevant, since the apply takes 350M, but it's an interesting (and very simple) trick. Making apply more memory-efficient is an interesting idea, but may require parsing the patch bundle twice.
David On 5/22/08, Reinier Lamers <[EMAIL PROTECTED]> wrote: > Op Thursday 22 May 2008 13:47:07 schreef David Roundy: >> On Thu, May 22, 2008 at 04:25:05AM -0700, David Roundy wrote: >> > On Thu, May 22, 2008 at 4:22 AM, Reinier Lamers <[EMAIL PROTECTED]> > wrote: >> > >> Did you actually test that this affects the memory use of put? It'd >> > >> greatly surprise me if it affected our memory use by more than a >> > >> couple of percent, and for such a small gain, I'm not convinced it's >> > >> worth the complexity. Of course, it could also either improve or >> > >> hurt >> > >> the cpu time used in put and apply, and I'd be interested in knowing >> > >> which effect it has. >> > > >> > > On my box, the memory usage went from 1 gigabyte and counting to 64 MB >> > > for the particular phase in which it consumed too much memory, and the >> > > maximum memory usage during the whole "darcs put" went from more than >> > > one gig to about 200 MB. >> But I can't help but wonder, did >> you run your testing with profiling enabled, by any chance? > > Indeed, I can only reproduce the abysmal performance of the code without my > patch when I turn on profiling. Otherwise it performs just fine. I learned a > lesson here. > > And my patch does not solve the bug reporter's problem: on a repository of > similar size, darcs also uses lots and lots of memory with my patch. > > Reinier > > > _______________________________________________ > darcs-users mailing list > [email protected] > http://lists.osuosl.org/mailman/listinfo/darcs-users > _______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
