NVIDIA releases CUDA 5.5 supports ARM
A few minutes Nvidia surprised the world by announcing that his unit moved to the business model GPUs Intellectual Property “IP” , modality with which license its graphics architectures to various companies, including expected many arm soc manufacturers.
To complement its IP graphics architectures, Nvidia has released the new version 5.5 of the API gpu accelerated computing (GPGPU) owner CUDA, which among its major developments expected brings official support for ARM microprocessors, allowing future SoC ARM-based tegra Kepler as 5 “Logan” and similar designed by other companies that license designs Nvidia CUDA applications to run.
CUDA 5.5 also brings some new features for developers in order to optimize the performance of their applications such as profiling, analysis tools, and other tools to identify and resolve bottlenecks that limit performance.
Nvidia also shows the Neural Network project at Stanford University, dedicated to investigating how the human brain learns, project that uses 16 cGPUs servers equipped with Tesla, which are able to model a neural network with 11.2 billion parameters.
Another project mentioned is Nuance company known for its speech recognition solutions, the tpecnicas using neural networks to improve training techniques in the automatic learning of the user’s speech peculiarities, so that these take the shortest possible time.
Link: NVIDIA @ ISC 2013: CUDA 5.5 Released & More (Anandtech)
.Tags: 5.5, accelerated, acelerado, API, arm, computing, computo, CUDA 5.5, GPGPU, gpu, logan, Neuronal, Nuance, Nvidia, soc, standford, tegra, Tesla