On Fri, 09 Dec 2005 12:30:46 +0100, Pete Vickers wrote: > [EMAIL PROTECTED] /tmp> dd if=/dev/zero of=junk.data bs=1m count=500 > 500+0 records in > 500+0 records out > 524288000 bytes transferred in 68.774 secs (7623327 bytes/sec) > [EMAIL PROTECTED] /tmp> > > [EMAIL PROTECTED] /tmp> dd if=/dev/zero of=junk.data bs=512k count=1000 > 1000+0 records in > 1000+0 records out > 524288000 bytes transferred in 68.576 secs (7645311 bytes/sec) > [EMAIL PROTECTED] /tmp>
Interesting ! - Just for the discussion, here are those on the ML350; running bsd: # dd if=/dev/rsd0a of=/dev/null bs=1m count=1000 149+1 records in 149+1 records out 157100544 bytes transferred in 1.690 secs (92944108 bytes/sec) # top # dd if=/dev/zero of=junk.data bs=1m count=500 500+0 records in 500+0 records out 524288000 bytes transferred in 6.655 secs (78778723 bytes/sec) # dd if=/dev/zero of=junk.data bs=512k count=1000 1000+0 records in 1000+0 records out 524288000 bytes transferred in 6.540 secs (80165466 bytes/sec) All with negligible CPU overhead. My main problem remains the very wrong timing of the bsd.mp kernel as pointed out in another thread. I made a speed comparison with both the supplied kernels, and found the dual CPU times slightly above the single times (97 versus 92) for raw data; but for files, the speed of dual CPU is similarly below the speed of single CPU. (Everything else identical; only boot to bsd / run tests / reboot to bsd.mp / run tests.) Uwe
