[poisondwarf]

Motherboard P4C800-e Dlx, WDigital SATA boot drive on Intel controller and second Maxtor SATA on Intel Controller. One Samsung PATA on Intel Controller. Two ATAPI devices as master and slave on remaining Intel Controller. One Maxtor USB (2.0) drive for external backup.

This arrangement allowed problem free back up to USB drive using either Ghost 9 or True Image 9.

Then I added a Maxtor SATA to the Promise Controller (in IDE mode) and went through the usual problem of "device cannot start" a problem solved through thread on TSF which showed what was necessary to install Promise drivers. Once this was solved the following problem emerged and this is the problem for which I'm seeking help.

All drives work well, including the USB drive, under all tested conditions except when I try to use Ghost or TrueImage to do USB backups both programmes fail due to a "...delayed write error" in Windows.

I have tested the problem in multiple ways that include disabling the "tagged queuing" and "synchronous transfers" on the SCSI properties of the Maxtor on the Promise Controller. No matter what permutations (or expletions) I try the results are consistent, neither back up programme works to the USB drive any more.

Both programmes do, however, work to back up my boot drive (say) on anything except the USB. Clearly, there is something going on that is beyond my ken and if there's anyone out there with a suggestion or a solution I'd love to hear about it!
PD

 

[poisondwarf]

Reply to self (since not know how edit post!)

Well, I finally got time to try the obvious. Disabled the Promise IDE controller from Device Manager. Now image programmes work fine once more in backing up to USB external drive. I assume that the PCI bus must be involved in the "delayed write" and "data lost" messages that come when the Promise controller is enabled.

This is one way to fix the problem but I'd welcome:

1. Any help to understand the problem and possible role of the PCI bus​

2. A fix that doesn't involve disabling the controller​

PD

 

[please]

@ poisondwarf
I don't know for sure but since the Promise comtroller and USB ports share the PCI bus, there may be a flow of data conflict there, like reading and writing from master to slave on the same IDE channel similar to a buffer underrun.
Another thought, according to the BIOS Optimization Guide from Adrian's Rojak Pot, I wonder if lowering your PCI Latency might help free up PCI bandwidth faster.
A third thought, try looking at your IRQs and see if there is a sharing/conflict with USB and Promise controller ports...try plugging into different usb ports.

 

[poisondwarf]

thanks for suggestions

@ poisondwarf
I don't know for sure but since the Promise comtroller and USB ports share the PCI bus, there may be a flow of data conflict there, like reading and writing from master to slave on the same IDE channel similar to a buffer underrun.
Another thought, according to the BIOS Optimization Guide from Adrian's Rojak Pot, I wonder if lowering your PCI Latency might help free up PCI bandwidth faster.
A third thought, try looking at your IRQs and see if there is a sharing/conflict with USB and Promise controller ports...try plugging into different usb ports.

thanks for these suggestions. I do know that there are no reported resource or sharing conflicts reported but I will try changing the USB port coonection and look into experimenting with the PCI latency.
PD

 

[poisondwarf]

Success: Disabled Spread Spectrum

Thanks to your suggestion, Please, and to your link to Rojak's website I was able to do some browsing and virtual navel gazing. I saw some references to spread spectrum in various sections (CPU, AGP, IDE and so on). A common theme in these references was to stability and timing problems, particularly if one is overclocking (and I am).

Looking in the BIOS for the Asus P4C800-E, I saw one reference to Spread Spectrum in the Chipset section and noticed the default setting is Enabled. I disabled it and then tried the backup routines. They now work!

The compulsive part of me wants to know exactly why this is so but the practical part gladly accepts the result. I wonder if the information would help some of those people who report problems writing to USB drives.

PD

 

[please]

@ poisondwarf
Hey great news! Thanks for letting us know what worked for you. That guide is really something nice.
I also have my Spread Spectrum disabled (from default enabled) for OCing, but I am a bit surprised to see it affected your system in that way...live and learn