Disappointing SCSI performance

[4gtek]

I'm using a Seagate Cheetah 18 gig SCSI U320 15K hard drive in my desktop system (P-4, 2.8, 512 meg ram, Win XP, LSI U320 SCSI controller) and have been very satisfied with the performance. Programs open very fast, everything is nice.

I'm running out of space on the hard drive, so I purchased the same drive, but 36 gig, same specs. I ghosted the old drive to the new one and swapped out the drives, but the performance is NOTICEABLY slower!

The system takes more than TWICE as long to boot; programs that used to open in less than 10 seconds now take 30, etc.

The model number of the new 36 gig drive is: ST336754LW. I'm not sure of the model number of the existing 18 gig drive, but it's similiar. If it's needed, I'll shut down and pull it to check it.

If anyone has any ideas as to why this new drive is so much slower than the old one, I'd appreciate it.

 

[smah]

Make sure that you have installed the current firmware for both the drive & controller.

 

[4gtek]

Thanks for your response.
I checked out both LSI and Seagate's sites for newer firmware.

I have the latest for the LSI controller, and there doesn't appear to be any for the drive. (that I saw, anyway)

 

[rvnguy]

Do you have SP2 installed??

See this link...does not address the disserence issue though

http://www.seagate.com/support/kb/disc/faq/winxp_scsi_slower.html

Double check LUN, device # assignment, new drive jumper settimgs. It should error if you have conflicting settings but never hurts to check

rvnguy
"I know everything..I just can't remember it all

 

[4gtek]

I do have SP2 installed. Also, I'm using FAT32, but I don't think that should make difference as the original drive is 18 gig, and the new drive is 36 gig.

I Ghosted the old drive over to the new drive, and replaced it in the system. When I saw the performance difference, I took it out and put the old drive back in.

The new drive was configured exactly the same as the old drive, with the exception of the SCSI ID number.

Is it possible that this drive is just not as good (performance-wise) as the older model? According to the specs, the 36 gig should be just a tiny bit faster.

(compare specs here):

http://www.seagate.com/support/disc/specs/scsi/st336754lw.html

http://www.seagate.com/support/disc/specs/scsi/st318453lw.html

36 Gig-Model # ST336754LW
18 Gig-Model # ST318453LW

 

[Freestone]

Have you checked in Event Viewer to see if there is any type of SCSI errors or informational messages?

 

[rvnguy]

but the performance is NOTICEABLY slower

If by this you mean real difference, you can produce some benchmark results. i.e. Google for things like 'hd benchmark' etc. and validate your suspicions. I would then contact Seagate and forward files produced indicating the problem. Seagate should fix this as this is a new drive.

before this, you might try setting to same SCSI ID as the old, as you are replacing your old drive(can't have both installed same time) in the event there is some pointer?? in the OS as to ID# and you have 'ghosted' all this information.

http://www.simplisoftware.com/Public/index.php?request=HdTach

http://www.freshdevices.com/freshdiag.html

http://active-hardware.com/english/benchmarks/benchmarks.htm

ATTO benchmark $

http://www.attotech.com/software/app1.html

Hope this helps.

rvnguy
"I know everything..I just can't remember it all

 

[4gtek]

Thanks for the suggestions.

I've already sent a note to Seagate-We'll see if they have any response.

In the meantime, I'll explore your benchmarking suggestions during the week and see how things stack up.

But, there's no denying that even without benchmarking utilities, it's very obvious: 35 seconds to boot (from power on) with the old drive and almost 80 seconds with the new drive. Also, Quickbooks 2005 opens in about 14 seconds on the old drive and it takes over 25 on the new drive.

 

[rvnguy]

Did you try setting the new drive to the SCSI ID that the old was set to???

rvnguy
"I know everything..I just can't remember it all

 

[4gtek]

Actually, I haven't had a chance yet, but I'll try that first, and also look at the benchmarking utilities you suggested.

It wouldn't seem that the ID would have anything to do with it, but it's a simple thing to try.

Thanks for the suggestion, and I'll let you know how I make out.

 

[rvnguy]

It wouldn't seem that the ID would have anything to do with it

I agree, but there might be some registry references as to where things are by ID or not.

rvnguy
"I know everything..I just can't remember it all

 

[4gtek]

I've had a response from Seagate Tech support, and they sent me a new firmware update for the drive, which solved the problem.

Thanks to SMAH, who correctly suggested this route right off the bat, but I guess I didn't look hard enough on Seagate's site for the update.

And thanks for the other suggestions as well. In the course of my investigations, I realized another machine I had built using a SCSI drive did not have a terminator on it. (mine did). I just forgot all about that when I built it. I changed the cable to one that had a terminator, but as it turns out, didn't seem to affect the operation or performance of the PC at all.

So now I'm wondering what all the fuss is about when it comes to termination. Also, if it IS essential, can I reuse the cable I was using (3 connectors), and just add a terminator to the far end? (and use the middle connector for the drive) But as I said, the PC functioned just fine, and there was no performance gain when I changed cables to one with a terminator at the far end-just past the connector.

Thanks.

 

[rvnguy]

Termination is totally important. Many devices produced today are self terminating, probably to assist with this nagging problem. Not all of them though, so it is still an item that causes sleepleness nights.

Check sepcs of the device. A SCSI chain must be terminated on both ends, normally the host(controller) is one end and includes termination.

So, being it is fixed, is it performing as expected?

Also Thank you for posting back with your solution as others can benefit from your experience.

rvnguy
"I know everything..I just can't remember it all

 

[4gtek]

Hopefully I didn't muddle things by mentioning the termination issue in the same post as the poor SCSI performance-they were not related.

The poor SCSI performance was TOTALLY solved by the firmware update on my personal machine.

The termination issue was on a customer's machine which I built recetly. (Mine had the termination all along)

During the course of investigating MY performance issue, it dawned on me that I had used a cable on the customers machine that had no termaniator. I did a "time to boot" test, as well as time to open an application (Quickbooks) and then replaced the non-terminated cable with one that had a terminator. There was NO difference seen in boot up time or application start.

Inasmuch as the PC was working perfectly WITHOUT termination, I wondered why it was needed.

Thanks.

P.S. The customers machine has the same SEAGATE SCSI drive as I just updated so I expect to do its firmware as well.

 

[edfair]

There are combinations of SCSI controllers and SCSI drives that seem to be immune to the extraneous noise generated by lack of termination. Then you go in the box and move the cable and all of a sudden nothing works, or worse yet, it get very intermittent.
So yes, it is important.
I've generally had good luck using drive based termination but others I have talked with insist that active termination is the only way to go.

Ed Fair
Give the wrong symptoms, get the wrong solutions.