Intel's Linux Sandy Bridge Graphics Still Troubling

The Intel Linux drivers were installed per their recommended Q4'2010 package versions, but barely anything works correctly. The next step was to pull from the Mesa 7.10 branch rather than master, in case something happened to break in the very latest Intel Mesa code. However, after doing so, everything was still malfunctioning. After that, I decided to try out the Linux 2.6.38 kernel, which is still pre-RC1 with the merge window not even having been closed yet, but it does bring a number of Direct Rendering Manager changes. While building the kernel though another Sandy Bridge issue was discovered: very high CPU load on the Linux 2.6.37 kernel with the Core i5 2500K will cause the system to lock up with a corrupted screen as if the GPU is being starved. It happened several times and within two minutes of beginning the kernel build process commencing.

Rather than not building the Linux kernel with multiple threads (thereby reducing the CPU load and hopefully avoiding the issue) where the Sandy Bridge's performance could not be exploited, I shut down the display server and just built the kernel from a console. It went a bit further without the X Server running, but even there, the Intel Sandy Bridge system again consistently failed. This though at least shows the problem is likely within the Intel DRM module. The CPU temperature was also checked at various times and this is not a thermal problem.