Agenda

e-Science with Many-Core CPU/GPU Processors

Schedule

The Summer School programme covers 5 days, from the 14th to the 18th of June, 2010.

The programme covers presentations from world renowed speakers, case study presentations, and hands on classes where participants will have the opportunity to work with CUDA. This hand on classes will cover both the single computer, and cluster programming environments.

Monday 14th June :

• 09h30 - 10h15 Reception
• 10h15 - 10h30 Welcome by Alberto Proenca and Keshav Pingali
• 10h30 - 12h30 Beginner's Tutorial on Many-Core Processors, Multi-Core Processors, and C Programming, Michael Garland, NVIDIA (slides)
• 12h30 - 14h00 Lunch
• 14h00 - 15h30 Introduction to Many-Core Processors, David Kirk, NVIDIA (slides)
• 15h30 - 16h00 Rapid-fire session (each participant presents himself)
• 16h00 - 16h30 Break
• 16h30 - 18h00 Introduction to CUDA, Wen-mei Hwu, University of Illinois U-C (slides)

Tuesday 15th June:

• 09h00 - 11h00 CUDA Threading Model & CUDA Memory Model, David Kirk, NVIDIA (slides)
• 11h00 - 11h15 Break
• 11h15 - 12h30 Algorithm Design for Many-Core GPUs, Michael Garland, NVIDIA (slides)
• 12h30 - 14h00 Lunch
• 14h00 - 16h00 Hands-on lab classes (Problems, Qsub script, Makefile.mac)
• 16h00 - 16h30 Break
• 16h30 - 18h00 Multiple GPU's in a MPI cluster, Wen-mei Hwu, University of Illinois U-C (slides)

Wednesday 16th June:

• 09h00 - 11h00 CUDA Performance Considerations, Wen-mei Hwu, University of Illinois U-C (slides)
• 11h00 - 11h15 Break
• 11h15 - 13h00 CUDA and Higher-level Tools, Libraries and Software Resources, Michael Garland, NVIDIA (slides)
• 13h00 - 14h30 Lunch
• 14h30 - 16h30 Teacher track (to discuss how to best use the textbook, resource sites, and labs)
• 16h30 - 18h00 Leisure Time

Thursday 17th June:

• 09h00 - 10h00 Case Study 1: Dense Linear Algebra for Hybrid GPU-Multicore Systems, Marc Baboulin, Universidade de Coimbra (slides)
• 10h00 - 11h00 Case Study 2: The Power of Heterogeneous Computing in Future Generations of Computer Graphics Systems, David Kirk, NVIDIA (slides)
• 11h00 - 11h15 Break
• 11h15 - 12h00 Case Study 3, Part I: Calculation of Electrostatic Potential Using Direct Summation - Steps of Performance Optimization, Wen-mei Hwu, University of Illinois U-C (slides)
• 12h00 - 12h45 Case Study 3, Part II: From Direct Summation to Cut-off Binning - Achieving Data Scalability in a Massively Parallel Computation, Wen-mei Hwu, University of Illinois U-C (slides)
• 12h45 - 14h00 Lunch
• 14h00 - 16h00 Hands-on lab classes
• 16h00 - 16h30 Break
• 16h30 - 18h00 New Fermi and CUDA 3.0 features, David Kirk, NVIDIA (slides)

Friday 18th June:

• 09h00 - 10h00 Case Study 4: Molecular Simulation Strategies for Next-Generation Streaming Architectures, Erik Lindahl, University of Stockholm (slides)
• 10h00 - 11h00 Project proposals, Part 1 (short presentations about how attendees would CUDA-ize their applications, followed by discussion and suggestions)
• 11h00 - 11h15 Break
• 11h15 - 12h15 Project proposals, Part 2
• 12h15 - 12h30 Closing Session