Re: raw vs block speed difference?

["Aaron J. Grier" <agrier%poofygoof.com@localhost>]

On Fri, Jul 16, 2010 at 04:47:27AM +0000, Michael van Elst wrote:
> The raw device uses IO buffers as you specify and limited to MAXPHYS
> (64k).
> 
> The block device uses IO buffers the size of a filesystem block (as
> recorded in the disk label) or 2048 bytes if DIOCGPART failed.  There
> is also some overhead for maintaining the buffer cache, but you CPU is
> probably fast enough to hide this.

repeating the raw vs cooked test on a different machine, dual 2.4GHz
Xeon with Ultra160 aic7899 controller:

# uname -r
5.1_RC3
# dd if=/dev/rsd3d of=/dev/null bs=64k count=8192
8192+0 records in
8192+0 records out
536870912 bytes transferred in 9.986 secs (53762358 bytes/sec)
# newfs -b 64k sd3d
/dev/rsd3d: 17501.8MB (35843664 sectors) block size 65536, fragment size 8192
        using 6 cylinder groups of 2917.00MB, 46672 blks, 93184 inodes.
super-block backups (for fsck_ffs -b #) at:
128, 5974144, 11948160, 17922176, 23896192, 29870208,
# dd if=/dev/sd3d of=/dev/null bs=64k count=8192
8192+0 records in
8192+0 records out
536870912 bytes transferred in 16.305 secs (32926765 bytes/sec)

that amount of overhead seems excessive.  am I missing something?

-- 
  Aaron J. Grier | "Not your ordinary poofy goof." | 
agrier%poofygoof.com@localhost