AccelerEyes has introduced, LIBJACKET, a broad and fast C/C++ library for GPU computing. With over 500 C/C++ functions, LIBJACKET represents the largest GPU computing library in the world. This CUDA-based library integrates seamlessly in any application enabling optimized utilization of NVIDIA CUDA-capable GPUs, including powerful Tesla compute devices.
LIBJACKET features include functions common to image processing, signal processing, as well as basic and advanced math functions.
Also, LIBJACKET benefits from key core technology, such as a runtime and just-in-time compiler, which bring optimized outcomes to developers and eliminate the hassle often involved in leveraging parallel hardware. LIBJACKET includes a GPU matrix class in C/C++ along with simple rules for matrix indexing and manipulations.
To date, AccelerEyes has been exclusively known for Jacket, a MATLAB GPU computing platform reportedly used by thousands of programmers worldwide. With LIBJACKET, the same core GPU computing technology which makes Jacket so powerful is now available to C/C++ programmers.
“LIBJACKET represents a major step forward for AccelerEyes,” said James Malcolm, VP of Engineering at AccelerEyes. “With LIBJACKET, any software developer can now accelerate their own application in the same way that we’ve been accelerating MATLAB since 2007. Programming with LIBJACKET is really easy and dramatically lowers the programmability barriers that prevent many people from jumping into GPU computing.”
LIBJACKET is currently released as a free beta product and will continually be upgraded during the beta period. The company anticipates the first commercial release of LIBJACKET to occur sometime in early 2011.