Graphics Hypertransport watchdog timeout error after upgrading GPU and PSU

I've had some stability issues with my system since upgrading the GPU. At first, I think I had some issues with previous driver versions (for the 5770) clashing with the newer Catalyst drivers for the 6950. Logging in as users other than the user I'd installed the new drivers through resulted in graphical artefacts, followed by a BSOD.

I removed the drivers via control panel and then used the "DriverSweep" to clear any remnants of the Catalyst drivers, before re-installing the latest version (12.1). This resolved the problems caused when logging in, however a new problem has emerged:

After the machine is idle (usually after several hours), I return to it to find it back at the Windows log in screen. Checking the event viewer shows a BSOD caused the system to reboot (despite me having configured Windows to _not_ automatically reboot following a BSOD).

The error in the event viewer reads:

Hypertransport watchdog timeout error
bugcheck was 0x00000124

Googling around suggests this error is tied to hardware, which concerns me. However, considering the BSOD only occurs when the system is idling and not during a gaming session, or running the Furmark stress test, I'm wondering if another issue is to blame?

As the problem is so difficult to replicate, troubleshooting myself has been time-consuming, so I figured I would post in these forums to see if anyone had any experience with this sort of problem in the past?

In investigating the problem, I have updated all the system's drivers, the motherboard's BIOS and installed the Sapphire Trixx software to check the GPU's status. As the system is idling when the issue occurs, I did wonder if the BSOD was caused by the system attempting to go into standby, so I've disabled the system from attempting to sleep in Windows' Control Panel.

Another thing I've noticed: using the Trixx software, the GPU core voltage was at 0.9V - when the application itself only allows the core voltage to be dropped to 1.1V; perhaps this is causing the card not to respond to the host in a timely fashion?

There are a few more troubleshooting steps I wanted to take, before I considered returning the card (but I hoped you might be able to shed some light on the matter before I undertake them - they are drastic and time-consuming!):

- Change back PSU (had a 600W Enermax model previously).
- Swap the GPU back (to ensure there's no problem with the MOBO).
- Re-install Windows from scratch (to rule out a driver issue).

I hope you guys can provide some assistance; I'm really happy with the card's performance otherwise, especially after installing the Skyrim HD texture pack!


Machine Details:
Motherboard: Asus M4A770TD Pro
Motherboard Driver: 1.3.0.49
Motherboard Chipset: AMD 770
Motherboard BIOS: 2301
CPU: AMD Phenom x2 3.1GHz Black Edition
Memory: Kingston 4GB 1600MHz DDR3 Non-ECC CL9 DIMM XMP Memory (Kit of 2)
Memory Latency: Default (set mobo to default settings)
PSU: Corsair VX550W
Graphics: Sapphire Radeon HD 6950 2GB "Dirt 3 Edition"
Graphics Driver: AMD Catalyst Driver 12.1
Sound Card: Creative Audigy 4
Sound Driver: 6.0.1.1377
Operating system: Windows 7 64bit - SP 1

 

An update on my progress:

I have had the machine running without issue for a couple of days after I switched Windows to "High Performance" mode in the power section of control panel. Essentially, the display and machine are never put to sleep.

I did read somewhere that if there's not enough power to the GPU when idling it can cause the PC to think it's "disappeared", so I'm going to switch back to my old PSU at some point to see if the extra 50W makes any difference.

Alternatively, I could see if using Trixx to keep the GPU core voltage at 1.1V is an option.

 

Further update:

- I completely re-installed Windows. It was probably was about that time anyway; there were a tonne of applications installed that we weren't using any more anyway. As an aside; I was extremely impressed with how Steam just picked up all my game data once I had re-installed the client! The machine had been running perfectly for a couple of weeks; including several consecutive days without the reboot occurring. Until last Saturday, when it rebooted while simply steaming music from Last.fm.

- Since then, I've been unable to successfully replicate the issue. This is despite me leaving the machine on for several days, occasionally playing Skyrim or BF3 and with it left running overnight in the same state as the previous times it's crashed (idling while logged into my fiancée's account, streaming music via Last.fm).

- Intrigued by the possibility of a PSU issue, I used the Thermaltake Power Supply Calculator and found that it suggested the power consumption of my rig would be between 560W and 580W, depending on how many USB devices I had connected. The PSU I installed alongside my GPU upgrade was the Corsair VX550W, which I had used to replace an older 600W Enermax PSU. Slightly embarrassed that I didn't check the power requirements before I swapped this out; guess I was just excited about having a nice new component in the system!

I'm going to swap the PSU back and try my hardest to replicate the issue again; my fiancée will be at home next week, so it's the perfect opportunity to see if it occurs again! Here's hoping it doesn't... I wondered if this kind of issue I'm experiencing could be caused by a struggling PSU? I would have thought I would have experienced BSODs while gaming, or running load/stress testing tools, but they only seem to occur while idle.