NVIDIA CUDA Compiler Source Code is Open Source Now
Printer Print Email Email
Bangalore: NVIDIA, the favorite graphics card manufacturing company has today announced that it Open sourced the Source code of its latest NVIDIA CUDA LLVM-based compiler. This would help the researchers and software programmers, to get hold of enhanced and easy access to GPU through the existing set of programming languages; In addition, LLVM’s modular design has made it possible for it to work on various processor architectures. The LLVM-based CUDA compiler is available at free of cost through its CUDA Toolkit (v4.1), which is the latest version that has a strong architectural support for NVIDIA parallel GPUs. The opening up of Source code could mean allowing budding entrants of new programming methodologies and accelerating of software development process. Also, the CUDA compiler has much resemblance with Java as it creates a virtual machine on top of varied processor architectures. The ones who could benefit the most out of it are the AMD GPUs and Intel x86 CPUs. The GPUs have proven facts that they accentuate the processing power for certain types of applications in areas such as computational biology and cryptology, apart from being a general purpose parallel processor. CUDA is a platform developed by NVIDIA, which was also designed to use parallel computing, in order to speed up computer operations, and to reduce the workload for the CPU in running certain applications including non graphical types. The CUDA has eased the video file format conversions, reducing significant amount of time while converting from any video format to H.264 codec, MOV, M4V, FLV, MP4 and MKV, on a CUDA enabled GPU. The NVIDIA wanted to reach out to other graphics chip manufacturers, who were till now denied the access to its rich graphical APIs; also making CUDA completely hardware independent. The processors that are compatible with CUDA are Power CPUs, Multicore ARM chips (including support for future 64- bit implementations), Intel's upcoming Many Integrated Core (MIC) co-processor and Texas Instruments' new floating-point capable DSPs. Furthermore, according to NVIDIA, the Chinese government through the Chinese Ministry of Education is planning to introduce NVIDIA CUDA-based training programs across its various universities across China to enhance and groom the local expertise. A spokesperson from NVIDIA said "By releasing the source code to the CUDA compiler and internal representation (IR) format, NVIDIA is enabling researchers with more flexibility to map the CUDA programming model to other architectures, and furthering development of next-generation higher performance computing platforms."

siliconindia