Nvidia has released CUDA 5.5 this month to bring GPU-accelerated computing to ARM platforms for the first time.
CUDA (Compute Unified Device Architecture) is a GPU- accelerated parallel computing platform and programming model developed by Nvidia. ARM is agreed to be the world's fastest-growing processor ecosystem and is approximately 10 times larger than the x86 CPU-based market.
The company suggests that programmers using CUDA will be well served with power and tools to build applications for advanced science, engineering, mobile and high performance computing (HPC) applications on ARM — and also x86 CPU-based systems too.
In addition to native support for ARM platforms, CUDA 5.5 has enhanced Hyper-Q support — now supported across multiple MPI processes on all Linux systems.
Hyper-Q technology aims to "slashes CPU idle time" by allowing multiple CPU cores to simultaneously utilize a single Kepler GPU, thereby advancing programmability and efficiency.
There is also MPI workload prioritization here to allow application developers to prioritize CUDA streams on the critical path first, optimizing overall application runtime.
"New guided performance analysis in the shape of Visual Profiler and Nsight Eclipse Edition now walks developers step-by-step through the process of identifying performance bottlenecks and applying optimizations. Fast cross-compile on x86 reduces development time for large applications by enabling developers to compile ARM code on fast x86 processors, and transfer the compiled application to ARM," said the company.
CUDA 5.5 also offers a full suite of programming tools, GPU-accelerated math libraries, and documentation for both x86- and ARM-based platforms.
