"Charles Kerr" <[EMAIL PROTECTED]> posted [EMAIL PROTECTED], excerpted below, on Tue, 29 Aug 2006 13:16:45 -0500:
> Yesterday I gave 0.110 the a.b.m.startrek challenge -- it failed. > The 'get headers' task was doing things that were find for regular > groups, even large groups, but once you get up past 1,000,000 headers > the overhead snowballed. > > I gave bugzilla a longish brain dump on the things I found wrong. > See http://bugzilla.gnome.org/show_bug.cgi?id=353334 for more info. > > More importantly, it also has a patch to fix this in 0.111. I tested > the patch against both 0.110 and 0.14.9x CVS, and the patch won. > memory comparisons are also included in the bug report. I got around to applying the patch here yesterday, and while I've not seriously stress-tested it yet, I've already noted the difference in CPU usage. The contrast between 0.109 and 0.110 with this patch reminds me of the difference between trying to use composite rendering on xorg-7.0, vs. 7.1 and finally 7.1.1 after switching to EXA from XAA. 7.1.1 w/ EXA and composite uses little CPU and is actually useful. 7.0 w/ XAA and composite was nice eye candy, but used up way more CPU and had too much built-in delay to actually be used, at least on my Radeon 9200 w/ xorg drivers. (7.0 EXA was too crashy, 7.1.1 XAA still uses CPU and I doubt that will change.) So now I get to use /both/ fancy EXA/composite /and/ fancy pan, without undue strain or slowdown. =8^) -- Duncan - List replies preferred. No HTML msgs. "Every nonfree program has a lord, a master -- and if you use the program, he is your master." Richard Stallman _______________________________________________ Pan-users mailing list Pan-users@nongnu.org http://lists.nongnu.org/mailman/listinfo/pan-users
