mkfs.xfs tune performance options (sunit and swidth)

["Mike" <samba@xxxxxxxxxxx>]

We are creating a single large filesystem 7x72GB using SCSI 10k disks.
The system is 2x2.4 GHz and 4GB RAM. The application which will be
accessing those drives creates large files and gets accessed
sequentially (not database) with multi  reads and writes.

I configured the 7 disks using RAID0 (/etc/raidtab below) and created
xfs filesystem:

raiddev             /dev/md7
raid-level                  0
nr-raid-disks               7
chunk-size                  256k
persistent-superblock       1
nr-spare-disks              0
    device          /dev/sdd1
    raid-disk     0
    device          /dev/sde1
    raid-disk     1
    device          /dev/sdf1
    raid-disk     2
    device          /dev/sdg1
    raid-disk     3
    device          /dev/sdh1
    raid-disk     4
    device          /dev/sdi1
    raid-disk     5
    device          /dev/sdj1
    raid-disk     6


#mkfs -t xfs -V  -f -l internal,size=32768b -b size=4k /dev/md7

# xfs_info /data2
meta-data=/data2                 isize=256    agcount=119,
agsize=1048576 blks
data     =                       bsize=4096   blocks=124473216,
imaxpct=25
         =                       sunit=64     swidth=448 blks,
unwritten=0
naming   =version 2              bsize=4096  
log      =internal               bsize=4096   blocks=32768
realtime =none                   extsz=1835008 blocks=0, rtextents=0


My question is how can I tune this to perform better for large file
reads and writes. The RAID Chunk size is 256K. Shall I make it larger?
What should I use for sunit, swidth, su, or sw. sunit and swidth are
confusiong, how those values relate to stride option on ext2/3? Any
other considerations?

Thanks,

Mike