Artifacts can be from a dying card or from it overheating, check the temps and make sure the fans are turning, and the heat sink is clear of any debris.
My PC has an interesting issue that has been hard to pinpoint, but I know there are a few experts on here.
Basically, half the screen has odd 'rainbow' artifact intermittently in all 3D games. And I mean, exactly down the vertical centre; everything on the left has the strange rainbow artifact (that is most apparent when 3D objects are in motion, such as looking around in Skyrim), but renders everything otherwise okay. This only happens sometimes, but persists through all games until a complete restart (sometimes) resolves the issue when it happens.
I have 2 EVGA GTX 465s in SLI and two monitors, one an LED Asus, the other a 3D capable LCD Acer.
Things I've Tried:
- Under/overclock settings
- Each GTX 465 has been individually tested at room temperature with different monitors, cables, power leads, PCIe slots, and in varying master/slave configurations when tested together.
- Custom fan controls have been in place since install, set to kick in at a lower temperature than OEM. Enclosure has two unobstructed filtered fans with direct airflow to cards, with eight total fans in the machine. GPU temperature is monitored/alarmed and does not exceed 85c at full load.
- EVGA OC Scanner does not detect any artifact after running in fullscreen for 20 minutes, even while games have the half-screen rainbow-motion issue.
- Drivers have been cleanly rolled back to a release I know works, and cleanly updated to the latest release. OEM Bios has not been touched.
This issue never causes hard lock ups, restarts, or error logs. Just... half the screen turns into motion rainbow randomly and persists until restart (which sometimes does not resolve the issue either, thus why I was able to test so many things).
The power supply should also not be the issue, but who knows. It's 850W and the one thing about this PC I spent liberally on.
Help?
Artifacts can be from a dying card or from it overheating, check the temps and make sure the fans are turning, and the heat sink is clear of any debris.
I suspected temperature issues as well, but I've run the EVGA alarmed temperature/fan monitor since installing the second card and I set the fans to kick in earlier than OEM (I was concerned the cards being so close together would cause temperature issues). The enclosure also has an emphasis on airflow. It seems to be present on either card in either slot when tested one at a time with different power supply leads. I also tried alternate monitor cables. The issue will resolve for like, no reason at all after a restart and may not come back for days (or it may persist for days).
Could it be anything else? What are the chances both cards are failing with the same rare issue? Also, is there a term for the whole half-rainbow-motion-screen issue? Even after the issue appears and I run the OC scanner for artifact, it finds none.
The 465's were some beastly hot cards, I had the 470 and it was the same way.
Question, did you have 465's that could be unlocked to the 470 shaders on the old GF10* die? If you changed the bios, you would want to change it back
TDP on those cards is 104*c, they can handle temps decently, but check everything. Especially with 2 of those cards you may need to rework your airflow if it is a heating issue. Too much airflow without proper steps to making sure all paths are unobstructed can also be an issue. Are your wires dangling all over the case or did you get zip ties and put them together? Putting them together can really lessen turbulence and drop temps a couple of degrees.
Do you have these Overclocked and over volted by chance? Reset to stock if you do.
Play around with drivers also, doing clean installs each time.