I've semi-recently built a new Linux rig for A/V editing and the occasional gaming. On some games I've experienced consistent reboots with high graphics settings that, I believe, are due to PSU overloading and its protection kicking in. Even at maximum load, the PSU should be well enough to feed the system.
cases where I didn't have problems:
OpenArena @ max settings, 5760x1080
QuakeLive @ max settings, 5760x1080
Serious Sam 3 @ max settings, 1080p (5760x1080 works for a while, then goes back to 1080p)
All valve games @ max settings, 1080p
Trackmania (via wine) @ max settings, 1080p
All 2D games
GPU accelerated video encoding/playing, flash
General 3-monitor PC usage
SpectraBall @ anything above lowest setting/resolution
Tiny and Big @ anything above lowest setting/resolution
Bitcoin mining on GPU
I've spent time tweaking game settings and reading high and wide across the net about this and it all points to PSU overloading. Strangely enough there are cases which are more graphic intensive than some of the fail cases - I don't know what to
The VC is powered via two 6-pin connectors that are on a single 12V rail
Just before the reboots, CPU and GPU temperatures are well in their respective limits, no error logs and system isn't at full load.
Nvidia proprietary drivers 310.44
Unfortunately I haven't bought a Win7 copy yet (*nix user, I've never got into games till now) and can't test it under Windows.
How to track down the actual problem. What component fails? What should I do to remedy this?
1. Is there any point in trying swapping the two 6-pin conenctors to the VC, or trying the molex-to-6-pin cables that came with the VC?
2. Is this a VC power overdraw or PSU overload?
3. Is this constant testing (intentionally causing reboots) bad for the hardware
4. Can this be a problem in the drivers?
5. Any way to measure what power is actually drawn from the PSU, the 12V rail and the VC?
Thanks in advance!