Re: 1394 Performance

On Wednesday 04 February 2004 14:34, Mark Knecht wrote:
> > > 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
>
> Humm...interesting results. As I said I haven't done any of this
> in at least 18 months now, but the buffer-cached value looks
> really high, while the buffered disk read is probably about right
> in my experience. As I recollect, I got numbers in the 10-15MB/S
> range for my FAT32 drive under Linux, while I could connect the
> same drive to Windows and calculated much faster transfers by
> moving large files and measuring times.

I am not sure here what real value the buffered-cache results have
here myself. It would appear that the faster the system the higher
the number. But it really doesn't reflect anything on the
throughput of the equipment under test. Now perhaps on older
(slower) hardware where the max buffered cache would max out in
comparison to the device throughput max it could indicate the
system is inadequate to start with. Or, in the case of sufficient
system to perform that a hardware / configuration problem might
exist.

My notebook, P-III/1.2 GHz, provided 100 MB/s. This system yielded
984 MB/s, P4/2.4GHz "C". I can see where in comparison to this
case, almost 10 to 1 improvement, that a P120 or 486 based system
might never reach even the buffered-cached read performance to
achieve maximum of the device being tested. So to me, the
buffered-cache reading is more to indicate what the system is at
least capable of producing. Perhaps something like a no load
loopback test.

Please correct me if I am wrong on this.

> > 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.
>
> That in itself is interesting. My expectations have always been a
> bit dashed WRT performance.

Interesting indeed. Up to now I have only really worked with DVD+RW
media at 2.4x for data purposes only. It is only recently that I
have ventured into the higher speeds and DVD video applications.

>
> > 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?
>
> Humm....I had intended to do this, but I placed my new DVD drive
> in an Oxford 911-based case and it wasn't even recognized by the
> 1394 stack on my Gentoo box, so I installed it as an EIDE
> internal drive.
>

Thank you greatly for this comment. It made me think of the bridge
devices. My 4x DVD burner is installed in an older enclosure
probably designed prior to the 4x DVD technology. This drive is
being replaced, hopefully tonight. I will install the new Sony
drive in a newer enclosure that I have that supports USB2 as well.
I will report back what my results are.

It may be possible that the Oxford bridges have firmware internally.
I seem to recall an issue as well in regard to hard disk capacity
in excess of 132 GBytes as being a problem. I do not know if it is
the bridge itself that had the problem or the Windows drivers they
supply with the enclosures. But new products now usually advertise
if they support drives larger than 132 GBytes.

>From all the responses I have received at this point though, I have
concluded that my system is operating at comparable performance.
The system is capable of delivering double the required 5.5 MB/s to
burn a 4x DVD. So either the problem is the bridge device or the
drive firmware and media problem. The Optorite DD0203 has been
plaqued by firmware problems from what I can gather from the web
pages, even in Windows environment. That is why I am replacing it.
A new drive and newer enclosure should reveal if this is the actual
problem.

Thank you once again.
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 Thu Feb 5 04:36:29 2004