I'm curious whether Robert's patch might have an effect on samba3 performance as well.
This isn't very methodical, but I see about 60% increase samba3 read bandwidth (from 14MB/s to 26Mb/s) when changing from the default kern.timecounter.hardware=i8254 to TSC, which makes me wonder whether it might. Below are details.. Rich Murphey ------------------------------- Server: Samba3 on a dual xeon running FreeBSD 6.0 release. Client: nbench on windows XP SP2. Client and server are using intel pro 1000MT gigabit cards. Here's the nbench output w/comments: w/sysctl kern.timecounter.hardware=TSC: Disk Performance, MBytes/sec File size: 100.0 MBytes thread: 0 write: 7.93 read: 22.94 followed by kern.timecounter.hardware=i8254: Disk Performance, MBytes/sec File size: 100.0 MBytes thread: 0 write: 6.80 read: 14.01 and back to kern.timecounter.hardware=TSC: Disk Performance, MBytes/sec File size: 100.0 MBytes thread: 0 write: 7.77 read: 23.61 _______________________________________________ freebsd-performance@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-performance To unsubscribe, send any mail to "[EMAIL PROTECTED]"
