Part in question is MSI 6950 Twin Frozr II, new.
When starting a Furmark or OCCT type stress utility, 99% of the time the desktop (Win7 64) hangs and then delivers a successful-driver-restart message. Sometimes the stress utility proceeds but usually the process has to be terminated.
Interestingly, when the GPU does successfully enter the OCCT stress test, there are no artifacts, and the frame rate is good. Failed entry into the test can manifest as many artifacts in the first rendered frame.
I have actually unlocked the 6970 shaders and again, if the stress test actually starts, it does very well (no artifacts).
Drivers are 11.9. I have stopped short of reinstalling Win7 but have otherwise cleaned up the drivers and tried a reinstall. Turned off Aero for kicks -- no change.
Truth is I have been out of the game too long to know if this is a software or hardware thing. The Google provided the idea that the clocks are too low in 2D mode, and this somehow causes the 3D transition to fail. I'm not sure how to create a CCC profile to test the idea, and it sounds like working around a hardware problem anyway.
Will the collective [H] brain help me out?
When starting a Furmark or OCCT type stress utility, 99% of the time the desktop (Win7 64) hangs and then delivers a successful-driver-restart message. Sometimes the stress utility proceeds but usually the process has to be terminated.
Interestingly, when the GPU does successfully enter the OCCT stress test, there are no artifacts, and the frame rate is good. Failed entry into the test can manifest as many artifacts in the first rendered frame.
I have actually unlocked the 6970 shaders and again, if the stress test actually starts, it does very well (no artifacts).
Drivers are 11.9. I have stopped short of reinstalling Win7 but have otherwise cleaned up the drivers and tried a reinstall. Turned off Aero for kicks -- no change.
Truth is I have been out of the game too long to know if this is a software or hardware thing. The Google provided the idea that the clocks are too low in 2D mode, and this somehow causes the 3D transition to fail. I'm not sure how to create a CCC profile to test the idea, and it sounds like working around a hardware problem anyway.
Will the collective [H] brain help me out?