Home > TERATEC FORUM > Program > Workshop 4

TERATEC Forum 2011
Workshop 4
SOFTWARE TOOLS FOR HYBRIDES AND GPU ARCHITECTURES

Numerical simulation is vital for enterprise competitivity but the productivity gains required cannot be decoupled from economic reality. Recently the CPU-GPU association has emerged as an innovative, efficient and sustainable solution.

However questions remain, like:
•  How to efficiently port enterprise codes on these architectures?
• How to adapt algorithms and numerical methods to exploit the performance potential of CPU and GPU?
• What is the software development methods adapted to these architectures?

The deployment of hybrid parallelized applications should be part of a corporate strategy in which managers have complete control and understanding. These aspects was discussed during this seminar through interviews and feedbacks from technology suppliers, software vendors and users

Chairman : Jean-Christophe BARATAULT, NVIDIA

14h00

Corporate experience: PSA production codes port to CUDA
Benoît DESCHAMPS, PSA

Click HERE to download the slides

14h30

Research Lab experience: REX PROMES port code financed by CAPS-GENCI
Patrick VILAMAJO, Laboratoire PROMES

Click HERE to download the slides

15h00 

ISV Experience: Non-Linear solver development in CUDA, ISV strategy and customer benefits
Jean-Luc LACOME, IMPETUS AFEA

Click HERE to download the slides

15h30 Coffee break – Exhibition - Networking - Grand Hall
16h00

GPU Computing status 4 years after its launch by NVIDIA, planned evolution for the coming 5 years
Jean-Christophe BARATAULT , NVIDIA

Click HERE to download the slides

16h30 

Bull HPC Competency Center: Oil & Gas seismic processing code port
Mathieu DUBOIS et Gunter ROETH, BULL

Click HERE to download the slides

17h00 

Code port methodology and CAPEX-OPEX
François BODIN, CAPS ENTREPRISES

Click HERE to download the slides

17h30 Conclusion

 

 

© Ter@tec - All rights reserved - Lawful mention