CUDA++: A Template-Based Meta-Programming Model for GPUs
Nyalia Lui (Indiana University-Purdue University Institute)
Co-authors: James H. Hill (Indiana University-Purdue University Institute)
We present our current work on using template meta-programming to improve the CUDA parallel computing platform and programming model for GPU programming. Our current results show that further abstractions on the CUDA language can be made through reductions in complexity, increase in modularity, and better performance.