38th SPEEDUP Workshop on High-Performance Computing

This workshop's focus: Multicore computing and Parallel Languages

EPF Lausanne, Switzerland,
September 7 - 8 2009

Sponsored by: Speedup   COLAB   Uni Basel   EPFL   DELL

Tutorial on practical use of GPUs


The tutorial will take place in room BC07-BC08 at EPFL.

9:00 - 10:30
Session 1: GPU basics
Welcome and introduction (Dominik)
Why GPUs? (Robert)
Ready-to-use GPU libraries (Dominik)
GPU architecture (Dominik)
10:30 - 10:45
short break
10:45 - 11:30
Session 2a: CUDA basics
Introduction to NVIDIA CUDA (Christian)
CUDA Parallel Programming Model and C for CUDA (Christian)
11:30 - 12:15
Session 2b: CUDA tutorial "hands-on"
basic usage of the CUDA tool chain
porting a simple data-parallel operation to the CUDA architecture using C for CUDA
12:15 - 13:15
lunch break
13:15 - 14:15
Session 3a: Advanced CUDA
CUDA Toolkit (Christian)
Optimizing Performance (Christian)
CUDA Directions (Christian)
14:15 - 15:15
Session 3b: CUDA tutorial "hands-on"
practical experiments and basic important optimisation steps
performance experiments and evaluation
15:15 - 15:30
short break
15:30 - 16:30
Session 4: Scientific computing
Aspects of scientific computing on GPUs (Robert)
GPU cluster computing primer (Dominik)
16:30 - ...
Session 5: Wrap up and discussion
Solutions to the Hands-on