Originally Posted by elvanor
I am not experiencing X lockups but complete system freezes (or could this be the same? how can I tell?)
Either use the nvidia-bug-report.sh shell script when you reboot your computer or do it via ssh while the system is still active.
As a comparative, I'm having the same problem and reported it in another thread.
nvidia driver version 260.19.29. 2 Nvidia GTX 295s in SLI mode (also happens with 2 Nvidia GTX 275s in SLI mode). I've tried various kernels from 2.6.34-gentoo-r1, 2.6.34-ck-r3, 2.6.36-gentoo-r5, and 2.6.36-ck-r5.
And to answer the obvious question with regards to optimization flags, I'm not one of the ricers, I use "-O2 -march=native -mtune=native -pipe" on the amd64 arch with 2 Opteron 2378s (2 quad cores). (And before you go "what are those march/mtune options", check the GCC man page, they use the optimizations that work on your current architecture and are not dangerous flags by any stretch of the imagination - which is why they're set to native.)
I'd say this became unstable for me in the last month or so, which puts the driver version somewhere after the 195 series up to current. 195.36.31 flickers for me (where it didn't before in August). And all the 256 and 260 versions have oddly become unstable regardless of the cards I use.
This, to me, points squarely at the drivers. Indeed, there is a hint of it in my Xorg.0.log with the X server reporting a problem in the video drivers, but nothing in the backtrace as the system goes into a spinlock and never finishes the trace. I'm in the process of re-emerging with the -g flag to see if it can narrow down, before it crashes, where the issue is within the driver.