Re: 1394 Performance

James (and list),

> On Tuesday 03 February 2004 23:00, Mark Knecht wrote:

> >> On Tue, 2004-02-03 at 19:27, James Finnall wrote:

> >>> > The best perfomance
> >>> > I have been able to get using the "hdparm -t" command is about
> >>> > 15 MB/s.
> >>
> >> How about hdparm -tT ???

> root@Merlin:~# hdparm -tT /dev/sdb
> /dev/sdb:
> Timing buffer-cache reads: 128 MB in 0.13 seconds =984.62 MB/sec
> Timing buffered disk reads: 64 MB in 4.57 seconds = 14.00 MB/sec

These numbers are consistent with my setup; Lucent FW323 firewire
controller, 3 chains, 1 drive pr chain. The drives are WD 120G, running as a
raid5 array. They are connected via Oxford controllers, but I am not sure
which Oxford controller is used. The enclosures USB2.0 as an option, so I'm
guessing OXUF922 chips.

[root@leire src]# hdparm -Tt /dev/sda

/dev/sda:
  Timing buffer-cache reads: 128 MB in 0.39 seconds =328.21 MB/sec
  Timing buffered disk reads: 64 MB in 4.55 seconds = 14.07 MB/sec

and
[root@leire src]# hdparm -Tt /dev/md0

/dev/md0:
  Timing buffer-cache reads: 128 MB in 0.40 seconds =320.00 MB/sec
  Timing buffered disk reads: 64 MB in 3.28 seconds = 19.51 MB/sec

I'm not sure why my buffer cache read-speed is so much lower than yours, but
this may be a chipset/memory-thing(?); ViA KT333 chipset, Athlon XP 1700

Also, I'm getting multiple lines of
"ieee1394: unsolicited response packet received - no tlabel match" in dmesg
while running a bonnie against the raid. I'm pretty sure this does not
happen with a single drive or a dual-drive stripe set. Any thoughts on this?

Yours,
-S

> >> None the less, people have commented for a long time that the
> >> Linux 1394 implementation isn't up to the performance that
> >> Windows gets from the same drives. I haven't done any comparative
> >> testing in a long time (maybe 18 months) but when I did I was
> >> only getting about 40-50% on Linux what I was getting out of
> >> Windows in the very same drive.
> >>
> >> NOTE: testing the same partition meant testing both with FAT32
> >> which is admittedly not the best Linux can do, but why should
> >> Linux be *any* slower doing FAT32? I don't know...

> Sorry, but I do not have any (at least convenient way) to test the
> system under Windows for any comparison.

> The only comparison I have is that I was using this entire subsystem
> before I upgraded my system. I did not have any reason to even
> question the performance. It always appeared to perform like I
> expected.

> In that regard of before and now, I may be expecting more from the
> system than before as well. This issue has become noticable
> because I cannot burn a DVD at 4x speed. So then perhaps another
> question for the list in general might be, can anybody burn a 4x
> DVD on the firewire bus?

> James

-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
mailing list Linux1394-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux1394-user
Received on Wed Feb 4 22:52:37 2004