The CUDA-GDB debugging tools apparently require a very specific x-windows configuration process and I am still trying to work out the steps. A person must install the NVIDIA drivers to access the CUDA features of the video card, but CUDA-GDB requires X-Windows to stop in order to debug applications running on the card, which then prevents the effective use of the CUDA-GDB debugging tool under X-windows. (FYI - I am using Ubuntu 12.04.3 LTS with NVIDIA driver for Linux x64 version 331.49). Before installing the NVIDIA CUDA driver I was had X-windows running on two monitors, one connected to the motherboard and one connected to the NVIDIA card with CUDA capability. After the NVIDIA driver installation I can only see the NVIDIA adapter and the connected monitor. How have other people configured X-windows to allow use of the CUDA-GDB debugger program?
I'm assuming there is some way to do the initial X-windows installation with the default system driver and then install a portion of the NVIDIA CUDA driver to support the CUDA functionality, but not to run X-windows?
Please note, I am not asking for someone to help me write an xorg.conf file for my motherboard from scratch. I am asking what the protocol was to go from a working X-windows configuration with the default drivers, to a working configuration that allows use of the CUDA-GDB debugging tool with full access to the GPU, without messing up the ability to use X-windows in the process.
Thanks, - Michael