Archive notice: This article was originally published on December 10, 2008. Links and embedded videos are preserved as part of the historical record.
OpenCL is a new API that enables developers to harness the massively parallel computing power of the GPU. Support for OpenCL is another important step in the evolution of the GPU, giving NVIDIA developers an additional programming option.
NVIDIA initiated the GPU-computing revolution with the introduction of NVIDIA CUDA, its massively parallel ISA and hardware architecture. With more than 100 million CUDA-compatible NVIDIA GPUs, CUDA has enabled significant performance gains across a wide variety of applications. It also provides a broad installed base of systems for developing applications with OpenCL. With support for additional standard programming languages such as C, Java, Fortran and Python, only the CUDA architecture can provide developers with a comprehensive range of programming environments for GPU-based application development.
First introduced with the NVIDIA GeForce 8800 and now standard on all current NVIDIA GPUs, CUDA is the foundation of NVIDIA’s parallel-computing strategy. CUDA has received an extraordinary response from researchers and developers worldwide because CUDA applications can be accelerated by a factor of 20 to 200 compared with CPU solutions. The CUDA architecture is now found in numerous computer systems, from supercomputers and workstations to consumer PCs. More than 25,000 developers worldwide use CUDA technology.
“OpenCL is an important innovation for the computer industry, and NVIDIA is pleased to play an active role in establishing the new standard,” emphasized Manju Hegde, General Manager of CUDA at NVIDIA. “We also welcome Apple’s pioneering role here. Apple’s assessment that the GPU will play an essential role in consumer applications marks an important milestone in the history of computing.”
