As of version 3.26, the Triton Ocean SDK is now built using NVidia’s CUDA Toolkit version 6.5 instead of 6.0. CUDA is the technology Triton uses to accelerate its wave equations on NVidia graphics cards, by spreading that computation out among the thousands of cores on your GPU.
This is important for a couple of reasons:
- If you’re a licensed user of Triton and are building it from source, you’ll want to install NVidia’s CUDA Toolkit 6.5 on your build system now. The main benefit is that CUDA 6.5 will let you build Triton’s CUDA DLL under Visual Studio 2013 natively, without requiring Visual Studio 2012 to be installed as well.
- This means your end users must have driver version 340.29 or newer installed on their NVidia-based systems, in order for Triton to perform its best.
We try to maintain a balance between keeping Triton’s dependencies current, while preserving compatibility with older compilers and architectures. As such, moving up to CUDA 7.0 probably won’t happen anytime soon, as NVidia dropped support for 32-bit applications with 7.0.