Hardware lurkers - help me find a new video card!

[Kevin]

One thing I wonder about is if my power supply is at fault here. My 4890 requires two 6-pin power connectors, while the 6770 only required one. Is it possible that one of my PSU's 6-pin power connectors is faulty and wasn't supplying enough juice to the 6770? Is that a thing that might make a video card work fine unless it was doing actual 3D gaming stuff? The PSU is a Antec 500W.

I can't say that it's specifically one connector being bad, but yes lack of power can do exactly what you described. I've seen systems doing that, swapped power supplies without changing anything else and system ran perfect.

Drivers, which you checked, are the next most likely problem. For some reason cursor artifacts, especially with dual monitors, seem to be common with Radeon cards. I hadn't seen any for around a year with my 7870, but the newest drivers (Oct 12th download) re-introduced them. This time at least the cursor only tends to flake out on the 2nd monitor and not the primary. Of course the slow loading and other display issues, while still potentially a driver issue aren't really the same things as the cursor issue.

The other most likely issue that I've seen with that is not heat on the GPU, but a memory issue on the video card.

So I've been messing around with some games, and both the 6770 and my old 4890 are crashing under load and giving me the "VPU recover" screen. I've only been testing them in WoW, Minecraft, HL2 and Portal - so not very graphically intensive. The temps on both cards seem to be just fine. The 4890 cranks up to 82°C or so, and the 6770 goes to maybe 45°C. Both look to be within normal under load.

Can I now be sure that I have a faulty power supply? The exact PSU I have is an Antec 500W Earthwatts. Is it at all remotely possible that I might have a motherboard/cpu/ram issue instead?

I've put in for an RMA for the 6770, but since it might not be faulty I'll be charged a 15% restock fee - plus I'll be out the shipping costs. Should I just keep it?

I'd hate to buy another new component and not have it solve my problems.

It could be a MB or RAM issue, I really don't see it being a CPU issue. You might have a component on the MB that is overheating under load, you might have a RAM issue in a higher range that is only being used under load. Heat issues on RAM causing issues like this are rare. I believe http://www.memtest86.com/ is still a solid tool for testing memory (Crucial and OCZ used to take test results from it unquestioningly for RMA requests). You could probably grab some sort of benchmarking tool that focuses on productivity stuff and loop it for awhile to check stuff too. Those don't tend to stress video too much, or allow you to turn off the 3D or graphics intensive stuff. Though that doesn't fully rule out the PSU either.