Hsuan-Hsiu (Senshaw) Ou, Spencer Davis, Chaochao Zhang, and Hai Jiang
.NET Framework, GPU
Graphics Processing Unit (GPU) has been introduced to High Performance Computing (HPC) community. Even with NVIDIA’s CUDA and Khronos group’s OpenCL, programmability is still one of the major issues to utilize GPUs effectively. In this paper, a preprocessor tool HPC.NET is deployed to map applications to GPUs. Programmers in Microsoft .NET field are able to use their familiar languages for high performance computing on GPUs. Multiple .NET languages are supported and certain parallelizable regions are detected and converted into CUDA code automatically. Experiments have demonstrated the effectiveness of HPC.NET.
Important Links:
Go Back
