File corruption with WD800JB

[LiamC]

I am getting a weird error in WinXP SP2. I have three hard drives installed
WD800JB, SP1213N and 6Y120M0. The Samsung and Maxtor have one dynamic partition each whilst the WD has three basic partitions and is the boot drive. At odd intervals, the following event shows up in the Event Log

EVENT 55--ntfs

The file system structure on the disk is corrupt and unusable. Please run the chkdsk utility on the volume C:.

Usually it is the E: partition that shows as having an error.

Now this is where it gets weird.

CHKDSK reveals nothing in error. I've run the WD extended tests and it shows no errors either.

A couple of times when the system has been trying to come out of a low pwer state, this drive tries to spin up, tries, then you can hear it spin down again, followed by a click. Repeat until you power cycle the PC. When you do this however, the Samsung (same IDE channel, slave) is not recognised. Power cycle again.

Anybody have this happen to them? If the WD diag software shows no errors, how am I going to convince anyone thre is a problem?

Merc is going to love this...

 

[P5-133XL]

My first guess is that the machine has a RAM problem: Your HD cache is getting corrupted. Try running Memtest86+.

 

[LiamC]

Mark, in the words of a great man; You are a freakin' genius!

Thanks.

 

[time]

Good call, Mark. :aok:

Bill, could the RAM explain some of your other issues?

 

[LiamC]

Hmmm, I spoke too soon.

On the weekend, I swapped out the Corsair XMS LL for some KingMax DDR500. As the RAM is only running DDR460, I set Command Rate to T1 (default T2) and CAS to 2.5 (3.0). So when Mark mentioned this, I thought straight away that that was the problem. Bumped everything back to defaults, but the problems still arose. Two runs of MemTest x86 3.2 and memory doesn't seem to be the issue. :-?

So Time, the short answer is it doesn't appear to be so--but I'm losing confidence in my t/s abilities so I won't rule it out.

I have since gotten a fault that pointed to a dud cable, so I swapped that out, and plugged in a Promise Ultra 100 in case it was the controller. The Ultra threw an error in the event log first reboot after installation though. :x

I've cleared everything and I'll keep an :cyclop: on things...

 

[LiamC]

Hmmm, looks like the cable was it. It was a very tight squeeze to get the two drives connected because of there locations. Looks like the pressure damaged the cable or the connectors. It has been 36 hours and no errors in the Event Log. First case of a dud cable ever.

I have left the Promise card in as now each IDE device has its own channel and there is no stretching of the cable(s) to make things fit.

 

[time]

Out of curiosity, why didn't you use SATA drives?

 

[LiamC]

SATA does not have any measurable performance improvement over PATA.

PATA is cheaper.

Ever tried to install XP on a SATA drive without a floppy drive in the machine or a diskette with the driver?

SATA marketed itself as being faster (see point 1) and I can't stand marketing bullsh!t. My little protest.

I do have a SATA drive now (Maxtor), but it's not a boot drive. But I do notice that the power and data cables bump/rub right up against one another, making that side of the connection hard to push in correctly, I keep expecting the connection to work loose and either the power or data cable to fail.

I do like the SATA connectors (well the idea, as opposed to MAxtors implementation) though. SATA-II likes like it might actually be an improvement.

 

[Buck]

Ever tried to install XP on a SATA drive without a floppy drive in the machine or a diskette with the driver?

Yes, and it's super easy.

SATA-II likes like it might actually be an improvement.

Be careful with SATA II, there are compatibility issues between this new protocol and VIA chipsets.

SATA does not have any measurable performance improvement over PATA.

This is true.

I do prefer the SATA cable manageability compared to PATA.

 

[time]

Ever tried to install XP on a SATA drive without a floppy drive in the machine or a diskette with the driver?

Yes, and it's super easy.

Just to clarify, nForce 3 and nForce 4 controllers (not sure about nForce 2) don't require special drivers when installing Windows - assuming the BIOS implementation is okay. You mainly need them for SiL (or VIA?), and you can slipstream those.

Perhaps you should consider an nForce board to replace your Abit?

 

[freeborn]

The main benefit of serial ATA is for system designers. Native SATA requires much less circuit board real estate and far fewer traces, making board layout cheaper. Eventually this will allow for many more SATA ports as PATA is retired. Differential signaling is more complex than parallel but as more and more drives are built SATA should eventually be cheaper all around than PATA. From a performance stand point there is not much to be gained with SATA unless your access pattern would benefit from queing.

Free

 

[Handruin]

I do prefer the SATA cable manageability compared to PATA.

This is my primary reason for using anything SATA related. Dealing with PATA cables is a PITA.

 

[LiamC]

Cable question(s)

What is the maximum length of a SATA cable?

Also, does USB 2.0 require a different cable to USB 1.0?

 

[freeborn]

Cable question(s)

What is the maximum length of a SATA cable?

Also, does USB 2.0 require a different cable to USB 1.0?

Sata I cables have a maximum length of 1 meter if they comply with the specification.

I think 15 feet is the maximum without a transceiver for USB cables regardles of 1.0 2.0 specification.

Free

 

[time]

Yes, 5 meters. I've used USB 1.1 further than that, but I imagine you'd be pushing it with USB 2.0.

AFAIK, many cables designed for USB 1.1 work just fine with USB 2.0. My unscientific rule of thumb is to look at the cable thickness - you don't see many thin USB 2.0 cables.

 

[blakerwry]

USB 2.0 high speed surprisingly has longer distance specs than the USB 1 low/full speed specs.

 

[time]

No, only 'low speed' or subchannel USB is limited to 3 meters (eg. mouse etc).

USB.org FAQ

 

[LiamC]

The bug is back

Event ID: 9 followed by Even ID: 51

So it only happens on the WD, and it is controller and cable independent. It's got to be the drive. It doesn't happen to the Maxtor or the Samsung.

Now to see if I can convince WD to RMA.

 

[LiamC]

Swapped the WD over for a Samsung. The event ID happens to the Samsung now.

So, the event is controller independent--it will follow a disk around if moved to another controller.

The event happens to whatever PATA disk is connected--so it is unlikely to be a disk error.

Cable error is not the issue.

The event only happens after three or four days of uptime.

Once it starts, only a reboot will fix the issue, whereupon, three or four days must elapse for the issue to reoccur.

Reading through this thread

http://217.115.198.3/content/topic/12946/?o=20

seems to indicate a conflict with a/the SATA controller. Time to remove the SATA disks...

 

[time]

You must have read a different thread to me ...

Quite a few folks had problems with SiL controllers - there was a known problem with these that required motherboard firmware upgrades.

At least one solved his problem by changing his CD/DVD drive - which kind of makes sense. Have you tried this?

Please tell me you're not still overclocking.

 

[LiamC]

Thanks time, I missed that it was with a SiImg controller

Overclcocking went out the window weeks ago.

 

[LiamC]

Solved. Well the problem went away.

I re-installed from scratch--and was super careful with my installation/driver install. I did this after running up the X2.

After a couple of weeks, I tried overclocking again and have the system running at 2.25GHz (250MHz FSB) for about four weeks now.

What I think I did wrong somewhere was run the memory @ 400DDR & 2 DIMM's when S754 only supports DDR333 with 2 DIMM's with C0/CG steppings. This is the only thing I can think of. When rebuilding the machine I set everything to auto/flashed to a later BIOS, though come to think of it, it could just be the later BIOS. By not writing everything down, my memory is hazy, I can't be 100% certain.

I have custom memory settings at the mo in it's overclocked state--and I have been very careful to stay within spec.

Me