Invalid context in Optix Prime

Hi, I have been playing around with the included samples for optix prime (simplePrime and its variants) and I only recently noticed that they are by default RTP_CONTEXT_TYPE_CPU, except for MultiBuffering and MultiGpu variants. When I try to change this to RTP_CONTEXT_TYPE_CUDA, it throws an error saying either “‘Invalid context’ (1)” or “An error occured with error code 1 and message Invalid context”. The first occurs when I modify simplePrime, and the second occurs when I use simplePrimepp or simplePrimeppMultiBuffering. The multi-gpu variant throws a third error, “Error on line 129:invalid device ordinal”.

The content of these makes me think there is some problem with recognizing the card(s), but all the CUDA samples run just fine (device query shows both GPUs perfectly fine) and all the other optix samples seem to run as well, so I’m not sure what the issue could be.

Some more information about your system configuration would be helpful:
OS version, OS bitness, installed GPU(s), display driver, OptiX version, CUDA Toolkit version.

Right, sorry. Got so freaked out about this that I forgot to provide information about my system.

http://imgur.com/9EwIhuH.png is the full device query. If you can’t see it, the gist is that I have a tesla C2050 that I am targeting and the toolkit is v6.0. Although there are 2 GPUs, the only GPU I’m really targeting is the tesla (in the future we may do multiple teslas, though). I’m using driver version 331.62. Linux version is 3.13.0-27-generic, ubuntu. It’s a 64 bit processor. I am using optix version 3.6.0, the beta version that was put out a little while ago. My next step is to install 5.5 (I hear 6.0 has some compatibility issues) and try the new stable 3.6.0. I was just hoping maybe it was something simple and stupid I had missed.