Hi Jos,
One thing to note is the reference to the VIA chipset, if your machines have this you may not be able to fix the problem. This and many other reasons is why I stopped using the VIA chipset some years ago.
BTW the problems only start when you start to push the motherboard to its limits.

>That is very kind of you to take that time. I am going to send this info to the network guy straight away. Many thanks for that effort Tracy :) I'll report back what we find as we go.
>
>
>>If new cabling and equipment doesn't resolve the issue, you may want to verify that you don't need this hotfix:
>>
>>http://support.microsoft.com/kb/321733
>>http://support.microsoft.com/kb/330174
>>
>>If you are running SP4 then it shouldn't be the issue.
>>
>>Some other possibilities:
>>http://searchwinit.techtarget.com/tip/0,289483,sid1_gci1041334,00.html
>>http://bermangraphics.com/problems/delayedwritefailure.htm
>>http://forums.viaarena.com/messageview.aspx?catid=23&threadid=50468&enterthread=y
>>http://www.thehindubusinessline.com/ew/2006/09/18/stories/2006091800270400.htm
>>
>>From the searchwinit link:
>>
>>Some common reasons for a delayed-write failure are:
>>
>>1. Problems with a device driver, especially a SCSI or RAID device driver. Some RAID device drivers are known to issue spurious "Delayed Write Failed" errors in XP Service Pack 2. Most manufacturers have been alerted to this, so check to make sure the disk drivers are up-to-date.
>>
>>2. Cabling problems. A faulty or broken cable -- especially for an external USB or Firewire enclosure—can generate this error. It can also happen if the cable is too long, or if it is hooked up through a hub that isn't up to spec. Another possible culprit is if you have a UDMA drive that requires an 80-pin cable, and you are using a 40-pin cable.
>>
>>3. SCSI termination errors. This has become less likely with the advent of self-terminating SCSI hardware, but it shouldn't be counted out.
>>
>>4. Media errors. This is the worst possible scenario -- essentially, drive failure. If you can garner statistics on the drive via SMART (such as SMART & Simple (http://www.beyondlogic.org/solutions/smart/smart.html), you may be able to determine if there's a mechanical failure in the offing. Gibson Research's SpinRite tool (http://grc.com/) is also useful for assessing media errors, but be warned: It may take a long time to do a thorough test.
>>
>>5. BIOS settings on the computer are forcing faster UDMA modes than the drive controller can handle. This is unlikely, especially with newer hardware (which can support UDMA far more flexibly), but it can usually be fixed with a BIOS upgrade, or by resetting the BIOS entries for the hard drives to auto-detect settings. Devices set to UDMA Mode 6 that produce this error, for instance, might need to be set to Mode 5.
>>
>>6. Controller issues. I've observed that USB controllers that contend strongly with other hardware can produce this error. In systems that have both "long" and "short" PCI slots (i.e., 64-bit and 32-bit), try moving the USB controller to the long slot. Older PCI cards will not fit in such a slot.
>>
>>7. Memory parity issues. If the problem appears after installing new memory, the memory in question may be faulty or not of the correct type for the motherboard in question. (This may go hand-in-hand with other problems such as random lockups, too.)
>>
>>8. The LargeSystemCache Registry tweak and ATI video adapters. One peculiar set of circumstances that has been observed on multiple machines with ATI video adapters and more than 512MB of memory involves the LargeSystemCache Registry setting, a DWORD entry found in HKEY_LOCAL_MACHINESYSTEMCurrentControlSetControlSession ManagerMemory Management. This setting governs the amount of memory set aside by the system for certain kernel processes. If it's set to 1 (which allegedly improves performance on systems with more than 512MB of memory), it can cause data corruption on some systems, and produce the "Delayed Write Failed" error. Try resetting it to 0 if it's been set.
>>