Tuesday, February 22, 2011

Installing CUDA.....

The Art of Multiprocessor Programming
There are several ways to install cuda on a machine having nvidia GPU

Requirements:

1.CUDA driver provided by nvidia
2.Cuda SDK provided by nVidia.
3.These steps are to install it in linux.
4.Download the nvcc package first (It's the nvidia c compiler)provided by nvidia

Why CUDA ?

Parallel Programming era has already arrived. With more and more types of processors coming up in the market, now is the turn of the softwares to exploit the resources provided by these processors.


Nvidia came up with an answer in the form of GPU having hundreds of cores and the CUDA architecture. Cuda is the world's first C compiler for the GPU.
Now one can think of using his personal computer with a GPU as a supercomputer.

CUDA - Programming the future

Cuda is short for Compute Unified Device Architecture. It is a parallel comouting architecture for nVidia's GPUs. It is a computing engine provided by nVidia in its GPU so that programmers can exploit the real power of massively parallel softwares.