Some of you may recall that a few weeks ago I shut down my quad core editing workstation as an energy-saving measure one Monday night. The following Tuesday, I started it up and Windows XP loaded up to a BSOD. After investigating the cause, I discovered all of the RAID devices were offline.

 

Fast forward a few weeks.. I bought a replacement motherboard, having concluded that a power spike may have damaged something, preventing the RAID controller from working when CPU Host Clock Control is enabled.

 

 

 

On Saturday, I installed that new motherboard. And... nothing changed! Same problem as before. So I was down to CPU, GPU and RAM. But then I started thinking about the PCI bus. The only thing that was failing on overclock was a PCI device--the RAID controller. What I could not understand was why the system worked for a year, then one morning it starts up and the RAID controller cannot load its BIOS anymore.

 

Well here's the strange part: That Tuesday morning, three weeks ago, unbeknown to me, the CMOS changed a few values on its own. It had changed the PCI e bus speed to 133MHz all by itself--I didn't touch that setting myself. The number *seemed* right and normal, so I never suspected anything.

 

But now I was completely out of options and started looking at stuff that I KNOW I did not touch and could not have caused to change. But somehow the CMOS changed the PCI bus speed on its own to 133MHz. So I started walking this speed downward, until I found the RAID reappeared at 120MHz. Now I am starting to wonder... is it 133 or 100 for the default? I start Googling for the answer. Some conflicting information suggests 133, some suggest 100. Some say 2.5GHz. Sheesh.. numbers all over the place!

 

So I concluded that the PCI e bus is supposed to clock at 100Mhz. How that became 133 at the flick of a power switch a few Tuesday mornings ago, I cannot imagine. But I did read some complaints of flakey behavior with F4 BIOS on this board. Others said that the flakiness went away with F7 BIOS. So I am now running F7 on the board and await to see how well it behaves from now on.

 

At least I nailed the problem. It feels good to be editing at 3.51GHz again. System is much snappier and I can handle unrendered effects playback smoothly once again.

 

Now I await the repaired Seasonic power supply that developed a weak +5V rail. It arrives on Sept 16. I'm running on a temporary and modified Seasonic from a single-core machine. It runs hot as hell, as it's only a 550W supply and is being loaded at or slightly beyond rated capacity.

 

And now I have that spare DQ6 motherboard... hmmm... maybe time to build a second quad core system...