FAQ

• What version of CUDA does the project currently support?

BrainGrid currently (as of July 2016) should support any version of CUDA as long as your GPU works with it.

• What types of graphics cards are supported?

Currently, only NVIDA cards that are CUDA compatible will work with BrainGrid. Check this page to see what devices currently support CUDA: https://developer.nvidia.com/cuda-gpus

• I don't have an NVIDIA card, can this run on AMD?

Unfortunately, BrainGrid currently only supports NVIDIA GPUs. You can still give it a test drive on your desktop, but the massive parallel computing speedups won't be realized unless you have an NVIDIA GPU. If we ever get around to supporting OpenCL, then AMD cards would be supported.

• Will you be supporting OpenCL in the future?

Who knows! The way this project has gone so far is that undergraduates or graduates will work on it in their spare time, with undergrads often doing a great deal of work for capstone projects. Almost the entire multi-GPU code was written by a single undergraduate as his capstone project. So if anyone ever wants to come in, branch off the repo, and add OpenCL support, more power to them!

• What Operating Systems does the project support?

Although we are hoping to get Windows support as some point, support is currently limited to Linux (probably any distro).