GPGPU 2014: Advanced Methods for Computing with CUDA
22nd -25th April 2014
Department
of Computer Science
University
of Cape Town
This workshop was a sequel to the successful GPGPU2013 workshop held at UCT in 2012. The principal lecturers on the course were again Nvidia fellows John Stone and Manuel Ujaldón, with additional lectures from UCT's Bruce Merry, Simon Perkins and Chris Laidler.
The workshop was attended by ~30 delegates, comprising postgraduates students (Honours/4th year, M.Sc and PhD) and researchers from academia and industry. The workshop covered more advanced methods and techniques suitable for programmers already familar with NVIDIA’s CUDA programming environment to enable effective general purpose computing with GPUs.
Speakers:
International:
John Stone
Associate Director, CUDA
Center of Excellence at University of Illinois at Urbana-Champaign
NVIDIA CUDA Fellow, 2010 to Present
Manuel Ujaldón
Associate Professor, University of
Malaga
NVIDIA CUDA Fellow, 2012 to Present
Local:
Bruce Merry
Dr. Bruce Merry completed his PhD in Computer Science at the University of Cape Town in 2007. From 2008 to 2011 he was a software engineer at ARM, developing drivers and middleware for low-power GPUs. During this time he represented ARM in the Khronos industry standards body (which defines OpenGL, OpenCL and other specifications). He is now a postdoctoral research fellow at UCT, specializing in general purpose computing on GPUs. He continues to contribute to Khronos and to the OpenCL specification.
Simon Perkins
Dr. Simon Perkins graduated with a PhD in Computer Science at the University of Cape Town in 2013. He is currently a Post Doctoral researcher in the department, working on GPU acceleration of the Radio Inteferometry Measurement Equation (RIME) for the Square Kilometre Array (SKA) project.
Chris Laidler
Chris Laider is in his third year of PhD studies into GPU accelerated blind searches for weak binary pulsars, using Dynamic Power Spectra and Hough Transformations.
Workshop
Organisers (Dept. Computer Science, UCT):
|
Chris Laidler |  |
Michelle Kuttel |
2014 Programme
| Tues 22 April | Wed 23 April | Thurs 24 April | Fri 25 April | |
|---|---|---|---|---|
| 8h30-10 am | John Stone (UIUC): GPU hardware overview |
Manuel Ujaldón (U. Malaga): New hardware features in Kepler, SMX and Tesla K40 |
John Stone (UIUC):Application Examples | "Hands-on in the cloud"" Manuel Ujaldón (U. Malaga) |
| 10-11am | TEA | TEA | TEA | TEA |
| 11-12:30am | John Stone (UIUC): Fundamental CUDA abstractions | John Stone (UIUC): Memory Systems | Manuel Ujaldón (U. Malaga: “New software features in CUDA 5.0, 5.5 and 6.0” and “Performance analysis based on the roofline model”
|
"Hands-on in the cloud" Manuel Ujaldón (U. Malaga) |
| 12:30-2pm | LUNCH | LUNCH | LUNCH | LUNCH |
| 2-3:30pm | Chris Laidler (UCT): Accelerating the Acceleration Search with GPUs | Bruce Merry (UCT):OpenCL | Simon Perkins (UCT):Parallelising the Radio Interferometry Measurement Equation (RIME) | |
| 3:30pm-4pm | TEA | TEA | TEA | |
| 4-5:30pm | Chris Laidler: tutorial | Bruce Merry: tutorial | Simon Perkins contd. |
 |
Funding for the workshop was provided by a South African Square Kilometre Array Project grant: The MeerKAT High Performance Computing (HPC) for Radio Astronomy Research Programme. |
