HeteroPar'2015 Program
Location
Room EI 3
Program
Session 1, Tuesday, Aug 25, 2015, 09:00 - 10:30
09:00-09:15 Meet and Greet
09:15-09:30 Welcome Note
09:30-10:30 Keynote
Session 2, Tuesday, Aug 25, 2015, 11:00 - 12:30
11:00-11:30 A randomized LU-based solver using GPU and Intel Xeon Phi accelerators, Marc Baboulin, Amal Khabou and Adrien Remy
11:30-12:00 Towards Community Detection on Heterogenous Platforms, Stijn Heldens, Ana Lucia Varbanescu, Arnau Prat-Pérez and Josep-Lluis Larriba-Pey
12:00-12:30 Accelerating Direction-Optimized Breadth First Search on Hybrid Architectures, Scott Sallinen, Abdullah Gharaibeh and Matei Ripeanu
Session 3, Tuesday, Aug 25, 2015, 14:00 - 16:00
14:00-14:30 Communication models insights meet simulations, Pierre-Francois Dutot, Millian Poquet and Denis Trystram
14:30-15:00 Modeling Contention and Mapping Effects in Multi-core Clusters, Juan-Antonio Rico-Gallego, Juan-Carlos Díaz-Martín and Alexey L. Lastovetsky
15:00-15:30 Identifying Optimization Opportunities within Kernel Execution in GPU Codes, Robert Lim, Allen Malony, Boyana Norris and Nick Chaimov
15:30-16:00 FiNS: A Framework for Accelerating Nested Simulations on Heterogeneous Platforms, Joris Cramwinckel, Ana Lucia Varbanescu and Stefan Singor
Session 4, Tuesday, Aug 25, 2015, 16:30 - 18:00
16:30-17:00 A Design Proposal for a Next Generation Scientific Software Framework, Anshu Dubey and Daniel Graves
17:00-18:00 Panel: Next Generation Heterogeneous Computing
- What are the main concepts to explore in Next Generation Heterogeneous Computing? What is about to happen?
- What are the main technical challenges to fix before Heterogeneous Computing can become commonplace? Will they be fixed before Next Generation Heterogeneous Computing?
- Are we addressing well the needs of the industry? What exciting applications will require Next Generation Heterogeneous Computing?
- How does Europe compete with USA, Japan, and China in Heterogeneous Computing? What are the concepts, technical ability, and infrastructure where it is lagging behind? How is it going to compete for the next decade?
Details about the peer-reviewed content:
We have accepted 8/26 submitted articles (31% acceptance ratio).
The list of accepted articles:
- A randomized LU-based solver using GPU and Intel Xeon Phi accelerators.
- Towards Community Detection on Heterogenous Platforms.
- Communication models insights meet simulations.
- Accelerating Direction-Optimized Breadth First Search on Hybrid Architectures.
- FiNS: A Framework for Accelerating Nested Simulations on Heterogeneous Platforms.
- A Design Proposal for a Next Generation Scientific Software Framework.
- Identifying Optimization Opportunities within Kernel Execution in GPU Codes.
- Modeling Contention and Mapping Effects in Multi-core Clusters.