DFI NF4 Ultra-D & onboard Via Firewire issue

[John]

Has anyone experienced file system corruption on external hard drives connected to the onboard via firewire port?

I have tried 3 different hard drives in 2 different 3.5" Bytecc usb 2/firewire enclosures w/ a prolific firewire controller. I can copy several GB's of data to the drive, then after an hour or two I can no longer access the drive unless I format it.

I am overclocking (217MHz memory bus / 260 HTT - passes memtest86+ 1.51, Prime95 torture Large FFT's for 20+ hours, SuperPI 32M), but I have read that the PCI bus is locked. I am using bios 0201. Right now I have it connected thru the USB 2.0 port and I'll see what happens. I may even try a NEC PCI firewire card as well.

Here are some of the errors in the XP Pro SP2 event viewer:

[*]Application popup: Windows - Delayed Write Failed : Windows was unable to save all the data for the file X:\$Mft. The data has been lost. This error may be caused by a failure of your computer hardware or network connection. Please try to save this file elsewhere.

[*]The system failed to flush data to the transaction log. Corruption may occur.
{Delayed Write Failed} Windows was unable to save all the data for the file . The data has been lost. This error may be caused by a failure of your computer hardware or network connection. Please try to save this file elsewhere.

[*]An error was detected on device \Device\Harddisk1\D during a paging operation.

[*]The file system structure on the disk is corrupt and unusable. Please run the chkdsk utility on the volume X:.

 

[John]

I hooked the external drive up to my Dell 700m via firewire and it works like a champ no matter how long it's hooked up. I just dropped in a Zonet ZFN2600 PCI firewire card w/ VIA chip into the bottom PCI slot on the Ultra-D, and I'll report back with the results later on.

 

[VirtualLarry]

Obvious question - have you repeated these tests at stock clock speeds? I'm not intimately familiar yet with that chipset, so if the firewire isn't bridged to the PCI bus, then it may still be overclocked. Similarly, chipset overheating could cause data-corruption issues. Finally, there are some issues with external firewire HD enclosures in Windows', period, which may be unrelated to the overclocking or any other hardware-level issues.

 

[John]

Actually I did not set my clocks to default to see how things worked out. I figured that since they should be locked, and I had no intentions of running my cpu at its default speed, I would try other troubleshooting steps to try and solve this strange issue.

The Zonet PCI Firewire card has solved the data corruption issue, and it allows my drive to have 0% cpu utilization.

 

[John]

Well I spoke too soon. I've had the drive turned on for 4+ hours now, and I just got a delayed write failed message. Now the file system is corrupted and my data is gone, lmao.

I guess at this point it is probably the prolific firewire chipset in the enclosure.

 

[VirtualLarry]

Or Windows. Or... what are the temps that your HD is showing, after extended operation in the enclosure? That could be an issue too.
Google "Windows firewire HD corruption"

click these for some more info

(PS. Definately read all the way through that last link. Loads of info there.)

 

[John]

Larry, thanks for taking the time to research this issue. Most of those posts point fingers at the Prolific 3507 chipset, which is what the Bytecc ME740U2F has, and that's the enclosure that I am using. The Seagate drive doesn't get all that hot, and the enclosure is barely warm to the touch. If anything I will get an enclosure with the Oxford 911 chipset.

However it looks like the "safest" bet is to stick with USB 2.0. Wouldn't you agree?

 

[Bar81]

Not really, firewire is still the superior option. If you can get a 1394b Oxford enclosure that would probably be best. Frankly I wasn't too impressed with the oxford 1394a/USB2.0 chipset.

 

[John]

Originally posted by: Bar81
Not really, firewire is still the superior option. If you can get a 1394b Oxford enclosure that would probably be best. Frankly I wasn't too impressed with the oxford 1394a/USB2.0 chipset.

Isn't a 64-bit PCI slot, or a mainboard w/ onboard FireWire 800 required for full speed?

 

[Bar81]

The 32-bit PCI bus does technically have enough bandwidth to support 1394b but since it's a shared bus you have to take into consideration any other devices connected to it and if you will be using them during transfer. Either way, you'll still be getting superior throughput to USB2.0. Here's a little more info:
http://www4.tomshardware.com/storage/20040402/firewire800-03.html