> > > Total time (wall clock) > > > orig: 5 hours + > > > no-reread: 6 minutes > > > > > Peak RES (as measured by top) > > > orig: 940MB > > > no-reread: 950MB > > > > Am I reading this correctly? This unoptimisation makes Darcs 50 times > > faster on large records while not using significantly more memory? > > At *least* 50 times faster -- Jason killed the first experiment after he was > satisfied that it was sufficiently bad on both time and space. :-)
Oh, and it also used significantly less memory. The RSS was approximately equal, but that is because Jason had only 1 GB physical RAM. The virt was significantly lower, and that might be important depending on the access patterns, as well as if your machine runs out of total memory including swap (which I have done many times with darcs + vmware + web browser + xemacs). Maybe next time someone runs this experiment they could look at page fault count... Regards, Zooko _______________________________________________ darcs-devel mailing list darcs-devel@darcs.net http://www.abridgegame.org/cgi-bin/mailman/listinfo/darcs-devel
