On Sun, Jan 20, 2013 at 10:09:43AM +1100, Dave Chinner wrote: > On Sat, Jan 19, 2013 at 01:08:47PM -0600, Mark Tinguely wrote: > > On 01/10/13 07:47, Jeff Liu wrote: > > I will have to go through this patch again and also test prints > > before and after the patch. > > > > Before the patch: > > write 108216 itrnc 219064 renam 305976 link 153144 remov 153144 > > symlk 158520 creat 157880 > > mkdir 157880 ifree 57912 ichng 1592 grwdt 44160 swrit 384 wrtid 384 > > addfk 69560 > > atriv 174720 attst 22456 attrr 87992 clagi 640 gwall 65024 grezr > > 4224 gwrfr 5760 > > > > > > After the patch: > > write 108216 itrnc 255928 renam 305976 link 153144 remov 153144 > > symlk 158520 creat 153784 > > mkdir 153784 ifree 57784 ichng 1592 grwdt 44160 swrit 384 wrtid 384 > > addfk 69560 > > atriv 174720 attst 22456 attrr 87992 clagi 640 gwall 65024 grezr > > 4224 gwrfr 5760 > > But it looks like you've already thought of that :) > > I'd also suggest that different geometries need to be checked, because > things like block size affect the result. I'd be looking at checking > these geometries: > > -b size=512 -n size=512 > -b size=512 -n size=4096 > -b size=512 -n size=65536 > -b size=4096 -n size=4096 > -b size=4096 -n size=65536 > -b size=4096 -n size=4096 -i size=2048 Just thinking aloud - probably a good idea to test different secotr sizes as well (-s size=4096) for a couple of those.... Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
