Nvidia releases CUDA compiler source code to public
On December 13th, NVIDIA announced it will open up the CUDA platform releasing source code for the CUDA Compiler to researchers and software-tool vendors. By opening source code, the company can easily add GPU support to more programming languages–further supporting CUDA applications on alternative processor architectures.
“The future of computing is heterogeneous, and the CUDA programming model provides a powerful way to maximize performance on many different types of processors, including AMD GPUs and Intel x86 CPUs,” in a statement by Sudhakar Yalamanchili, a professor at Georgia Institute of Technology.
With a modular LLVM, which is also enhanced with architecture support for Nvideia’s parallel GPUs, open source compiler infrastructure will make easier added support for new programming langues and processor architecture. The LLVM-based CUDA compiler will also include the latest release of of CUDA Toolkit version 4.1 and is available to the public. (available to qualified academic researchers and software tools developers)
“This initiative enables PGI to create native CUDA Fortran and OpenACC compilers that leverage the same device-level optimization technology used by NVIDIA CUDA C/C++,” said Doug Miles, director of The Portland Group. ”It will enable seamless debugging and profiling using existing tools, and allow PGI to focus on higher-level optimizations and language features.”