The instructions for cross compilation in this document: http://docs.nvidia.com/cuda/cuda-installation-guide-linux/#cross-platform are clear for the most part - I have some questions regarding it and would appreciate some feedback:
- Does the ‘host’ need to have the nvidia driver for the GPU in question installed?
- What exactly needs to be installed on the target (libraries? GPU driver?).
- If a driver is needed for the target, does the version between the host and the target need to match?
The host in my case is Ubuntu 64-bit 14.04 LTS.
The target is the Xilinx Zynq SoC (dual-core ARM9 - ARMv7l architecture) running Linux.
Thank you in advance.