[wizanon]

I've an Fujitsu 10krpm MAN3367MP SCSI drive on an adapTec 29160LP controller on a 32 bit PCI slot of an Nforce 4 (AMD) platform mobo, with an opteron and DDR400@2.5 cas

I have this result from HDTach, and it seems strange. The program won't even complete a long (32) test, it sites that an error has occured.



I also get a similar result (nicer graph perhaps) from the interesting Systool program which you can find on techpowerup, so I believe there is a real problem evident. The transfer rate drops more then I had expected across the 36GB in any case, so I may not use this drive for much longer.

Here is a composite of speedfan SMART analysis result display for the drive:



I don't know if it is likely that bad sector avoidance is now causing the dips at the start and just after the start of the original graph, which I'd like to eliminate.

Can anyone help me get optimum performance from this already fairly limited device?

Thank you.

[continuum]

Have you tried a different cable?

Are you trying HDTach with nothing else running in Windows?

[wizanon]

continuum, on Nov 20 2006, 07:11 PM, said:

Have you tried a different cable?

Are you trying HDTach with nothing else running in Windows?

I haven't got another cable to hand, unfortunately they are kinda expensive around here so I don't want to pick one up at the moment.

Windows (XP SP2.) was running basically bare and I ran the test a number of times with 2 different tools, and the dips in the graph were at roughly the same points ...

[continuum]

The dips in the graph make me wonder if another application is hitting the system (be it Windows, traffic on the PCI bus, etc.) right then, causing the dip.

[wizanon]

I found I was using the XP MS drivers (circa 2001), so I updated to 'Set 4 SP5 S4' drivers. The window in the background is Systool alpha 728.

The number of tests I am running (and their similar results) together with the background processes I tolerate (mostly I disable everything) really makes me suspect it is the drive showing a problem here. Has anyone seen this or something similar before? Did I have to pick between replacing the cable or the drive anyway, or might this be bad sector re-allocation causing seeks during the sequential read test maybe?

[wizanon]

Well, after reading U320 drive can't achieve higher than 62MB/sec burst rate, I used PCI Latency Tool (2 for win32) to set the latency of my adaptec Card to 248 and of my other two PCI devices, to 128.

Which seems to have fixed it! Whoop!



Unfortunately I didn't give you folks a chance to solve it in this thread so I'm sorry about that
Thanks for reading!