So I just installed three machines - a dual gtx470 and two dual gtx480s - with the latest driver, cuda runtime and SDK, and everything seems to have gone OK… X server still works and deviceQuery works, too. Tried to install an identical machine (except it has a gtx480 and a Tesla C2050) and not only does deviceQuery fail with some error such as bad device count or id (will clarify as needed), but booting the physical machine won’t get it into graphical mode (though remote access via ssh works normally).
I saw one other question a few down that seemed like somebody having a similar issue, but he said that it was a stupid question and removed it. It wouldn’t surprise me in the slightest if I’ve missed something stupid… can I not mix Tesla and non-Tesla? Does doing so require a special driver, runtime or SDK installation, or additional OS configuration? For what it’s worth, this is all running on top of CentOS 6.0 64-bit (based on RHEL6). Thanks again!