Based on a recent mailing list thread, I ran a few tests myself:
ad4: 238475MB <Hitachi HTS542525K9SA00 BBFOC33P> at ata2-master SATA150
[cmd=>]dd if=bsdanywhere44-amd64.iso of=/dev/null bs=3M[/cmd]*
sysct vfs.read_max=8
227+1 records in
227+1 records out
715751424 bytes transferred in 30.929360 secs (23141488 bytes/sec)
=16
21287127 bytes/sec
=32
51050018 bytes/sec
=64
47837291 bytes/sec
=128
49382747 bytes/sec
From the mailing list, the best value seems to vary a LOT based on hardware, mirroring, etc etc. "bs=" seems to have very little effect on the variances in speed.
* the .iso was different for each run to attempt to obviate this:
Results are unscientific, YMMV, HTH, GSTQ
ad4: 238475MB <Hitachi HTS542525K9SA00 BBFOC33P> at ata2-master SATA150
[cmd=>]dd if=bsdanywhere44-amd64.iso of=/dev/null bs=3M[/cmd]*
sysct vfs.read_max=8
227+1 records in
227+1 records out
715751424 bytes transferred in 30.929360 secs (23141488 bytes/sec)
=16
21287127 bytes/sec
=32
51050018 bytes/sec
=64
47837291 bytes/sec
=128
49382747 bytes/sec
From the mailing list, the best value seems to vary a LOT based on hardware, mirroring, etc etc. "bs=" seems to have very little effect on the variances in speed.
* the .iso was different for each run to attempt to obviate this:
Code:
> dd if=bsdanywhere44-amd64.iso of=/dev/null bs=3M
227+1 records in
227+1 records out
715751424 bytes transferred in 0.921832 secs (776444670 bytes/sec)Results are unscientific, YMMV, HTH, GSTQ