A while back Seth made an OpenCL enhanced version of FANN. This is still a separate download, but there are plans to include this GPU enhanced version in the main repository.
A while back Seth made an OpenCL enhanced version of FANN. This is still a separate download, but there are plans to include this GPU enhanced version in the main repository.
I’m doing a research project at my school, and was wondering when you guys plan to incorporate this into the official FANN code?
We are needing to train 10,000 – 60,000 data sets for our prototype, and having this pre-packaged with FANN would definitely make training easier =D
Are there plans to also have a CUDA enhanced version available which would be plug in compatible with existing applications? It would be great to be able to simply copy the new e.g. doublefann.c into my project and start using a Tesla to speed up both training and operation.