Page 1 of 1

Anyone see this before with a HD?

Posted: Fri Oct 22, 2004 6:25 pm
by Executioner
Image

This sawtooth. It's a SCSI drive connected to a Tekram DC-390U3W SCSI card. At boot, during the BIOS initialization, it reports that the drive supports 160mb/sec. When I installed XP, I did not load a driver since XP had one built in during the install, and there are NO drivers for XP on Tekram's site.

Anyone ever seen this before with a HD? My only logical conclusion is the drive has some problems at certain areas of the platter, or maybe the read cache is defective?

Just for kicks, I benched marked my other Seagate, a ST3120026A, 120gb IDE, and it's a gradual line that with no drops like the SCSI one.

Posted: Fri Oct 22, 2004 6:30 pm
by DoPeY5007
could it be CPU hang?

Posted: Fri Oct 22, 2004 7:31 pm
by smb
Actually, I read an article last year that said this particular controller had a problem with a "write thru flag" . As a matter of fact, I think I answered a GCS post as he was having a similar problem with an Adaptec controller as well. It's something that XP sets somewhere.

Posted: Sat Oct 23, 2004 12:18 am
by Executioner
Originally posted by smb
Actually, I read an article last year that said this particular controller had a problem with a "write thru flag" . As a matter of fact, I think I answered a GCS post as he was having a similar problem with an Adaptec controller as well. It's something that XP sets somewhere.
I did a search for "write thru flag", and some were able to fix the problem by changing their drives from basic to dynamic. So I decided to look into it on my system, but that option to convert is not available on my system. Some were saying that the 64 bit version of xp fixes the problem, but that sucks in my case, as I would have never even used a SCSI boot drive knowing this info.

EDIT:
OK, I found out why I did not have the option to convert to dynamic. I had to right-click on the drive number.

What is the difference between dynamic and basic? I have 3 HD's and all of them are setup as basic.

Posted: Sat Oct 23, 2004 1:43 am
by Executioner
Tried another bench program called Atto:
Image

After reading more on newsgroups, it seems that the xp bug was really a win2k bug, and it was fixed with sp3 in win2k, but I'm still not happy with the performance of a SCSI in xp pro.

Posted: Sat Oct 23, 2004 3:01 am
by DocSilly
OK, first lets take a look at my Seagate Cheetah X15 ST318451LW on the same controller under W2K SP4 (32mb test):

Image

Those numbers are comparable to the lowlevel results on Storagereviews test (28-42mb/s).

Atto only tests the sustained transfer rate and you left it at a total length of 4mb .... it should be at least 32mb to prevent it from being run within hdd cache only. Your hdtach would be comparable to Storagereviews lowlevel test (45-60mb/s) if they wouldn't drop to zero.

Now some more ideas how to "fix" your results since they really show that something is wrong here.
- Recheck the cabling, make sure you use a LVD cable with an active LVD terminator at the end of the cable.
- Update your SCSI BIOS, latest version is 4.19 on [url=ftp://ftp.tekram.com.cn/storage/SCSI/PCI/DC-390X/DC-390U3W/Bios/V419/]Tekrams FTP[/url] , I think one of the BIOS updates fixed some Seagate hdd related problems.
- Make sure you have no apps running that might access your hdd during the test, like p2p programs.
- You could also search the Storagereview forums.

Posted: Sat Oct 23, 2004 12:05 pm
by Executioner
I verifed the SCSI bios at boot, and it's version 4.19.0. I'm using a 68 pin SCSI cable with the LVD terminator on the cable, located after it connects to the hard drive. Here is a pic:
Image
So my SCSI ribbon cable comes from the controller card and attaches to the SCSI HD, then the next connector on the SCSI cable has the LVD terminator. Is this OK, or should it be located at the very end of the SCSI ribbon cable? The reason I picked the next open connector after the HD because I have no other devices on the cable.

Forgot to add:
I have nothing in the background running except for Notron AV 2002, HP Deskjet 932 icon, ATI control panel icon.

Posted: Sat Oct 23, 2004 12:48 pm
by DocSilly
Uh-oh ... terminator has to terminate the cable, so it always has to be at the end ;)
The terminator is the correct one ... and I had BT running in the background.

Posted: Sat Oct 23, 2004 11:09 pm
by Executioner
Originally posted by DocSilly
Uh-oh ... terminator has to terminate the cable, so it always has to be at the end
LOL thanks. That fixed it. The Tekram manual says nothing about having the LVD termination on the last connector on the cable. I think their manual can use some tweaking to better explain this. After moving the LVD terminatior, it's much more responsive when I open apps and such. Can't believe that's all it was.
Image

Posted: Sun Oct 24, 2004 2:46 am
by DocSilly
*cough* DC-390U3 Series User's Manual page 16 *cough* ;)
I'm glad it was so easy to fix.

Posted: Sun Oct 24, 2004 10:24 am
by Executioner
Originally posted by DocSilly
*cough* DC-390U3 Series User's Manual page 16 *cough* ;)
I'm glad it was so easy to fix.
LOL, yeah it's section 2.2.1, but I must have gotten confused, since I have never used this type of termination before. I have another system (old) in my garage running win2k with a SCSI boot drive, and the termination is done with jumpers. Simply hook up the cable and your set. Thanks again Doc for all your help. :)