CUDA

Legacy documentation This page describes a service provided by a retired ACENET system. Most ACENET services are currently provided by national systems, for which please visit https://docs.computecanada.ca.

Description

CUDA, from NVIDIA, is a parallel computing platform and programming model that harnesses the power of the graphics processing unit (GPU).

Modulefile

cuda

Commands

nvcc

Hardware/Software

NVidia Quadro FX 5600

NVidia driver installed is 319.37

CUDA Toolkit Version: 5.5

You can run deviceQuery to see the details.

Documentation

NVidia CUDA Zone

Tutorial at Google Code

SDK Code Sample

$CUDA/samples - SDK 5.5 code samples

$CUDA/sdk42 - SDK 4.2.9 that contains OpenCL code samples

Usage notes

The only ACENET machines equipped with GPUs are the four graphics workstations at Mahone. Consequently that's also where the CUDA compiler suite is installed, and nowhere else. Please use the Grid Engine to access these machines. For compiling and testing purposes you can request an interactive job, as in this example:

$ qrsh -l h_rt=3:0:0 -pe openmp 4 -q graphics.q -cwd bash
$ module load gcc cuda
$ nvcc source.cu -o executable
$ ./executable

For a batch job use

#$ -cwd
#$ -l h_rt=1:0:0
#$ -pe openmp 4
#$ -q graphics.q

module load gcc cuda
./executable