On Mon, 2002-03-18 at 21:12, civileme wrote: > Ext2 I ran three times under each mode and came up with the same results > (nearly). The one thsat others suggested to me as the fastest turned > out to be the slowest. > > Yes, i have run sledgehammer on IDE, on SCSI and on seven different > processors, including a PPro dual system running SMP. The verdict is > that ext3 is slow, about 60-70% of ext2, while Reiserfs, XFS and ext2 > are all comparable, and JFS is out in front except it takes more storage > and needs defragmentation as a separate process.
Your results with ext3 as compared against ext2 might explain somewhat why I haven't noticed alot of difference, although not totally. For one thing, I still don't know what ext3 journaling mode you were using for the tests. For another thing, I also still don't know what the minimum size for XFS filesystems are. 20 meg? 30? 40? I've looked, but have not been able to find that information anywhere. The reason I'm interested is because I'm considering using XFS when I migrate my configuration/data to LM82. Your tests have me interested. Concerning XFS minimum partition size...if you don't know, have you got an idea of where to go to find out? > Withint two weeks I will have my site back up and the timed version of > Filesystem Sledgehammer set up downloadable for everyone to try. The > only caveat is that only Mandrake kernels really run XFS as it should be > run, because XFS does require some significant kernel changes. Sounds like another feather in Mandrake's cap.I look forward to running the benchmarks on this system. > And sorry about the postscript. I was typing after a very long day. > > Civileme No problem. I am just grateful that *someone* is doing public filesystem tests. So a heartfelt "thanks" is in order. Best Regards, LX _________________________________________________________ Do You Yahoo!? Get your free @yahoo.com address at http://mail.yahoo.com
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
