Conferencia Final Supercomputación y eCiencia

La "Conferencia Final Supercomputación y eCiencia" tiene como objetivo presentar los resultados científicos de las diversas áreas científicas de este proyecto de investigación nacional.

Programa

Fecha: lunes 27 y martes 28 de Mayo de 2013

Ubicación:  Universitat Politècnica de Catalunya, Aula Master, Edificio A3

Lunes 27 de Mayo 2013

15.00 Introduction: Mateo Valero, SyeC Project Coordinator
15.15 Consolider Program: Marina Villegas, General Subdirector of Research Project of the Spanish Ministry
15.45 Keynote speech: “Multiscale Dataflow Computing”, Oskar Mencer
16.45 Coffee break
17.15 Scientific activities and results by area:
         17.15 Astrophysics
         17.45 Life Sciences
         18.15 Material Sciences
18.45 End meeting 1st day

Martes 28 de Mayo de 2013

09.00 Keynote speech: Future Strategies for Supercomputing”, Prof. Michael Resch
10.00 Scientific activities and results by area:
         10.00 Engineering
         10.30 Material Sciences
         11.00 Computer Sciences
11.30 Coffee break
12.00 Keynote speech: “The path to Exascale”, Karl Solchenbach
13.00 Keynote speech: "Energy Efficient Computing with GPUs", Timothy Lanfear (Nvidia)
14.00 Lunch
15.00 Conclusions. Mateo Valero, SyeC Project Coordinator
 

Charlas invitadas o Keynotes

Oskar Mencer
Senior Lecturer, Department of Computing, Imperial College London
Title:Multiscale Dataflow Computing

While on the chip level, microprocessors look like the most efficient way to compute, once we evaluate efficiency on the architecture level, the more flexible approach we employ with multiscale dataflow computing allows us to further minimize communication and balance computation with communication components. Finally, on the system level, parallel programming for both multicore and dataflow based systems requires significant effort, and legacy software presents a major obstacle to change. Performance results from users at universities and industry show the potential and typical advantage from multiscale dataflow computing.

Michael Resch
Director IT at University of Stuttgart
Title: Future Strategies for Supercomputing 

Abstract: Supercomputing has been a success story for more than 50 years. Driven by ever faster systems, ever better tools, and ever better numerical methods a variety of research fields has benefitted from this. Over the coming years the story will change. Limitations in hardware development and difficulties in developing scalable methods will create new challenges for the scientific community. This talk aims at working out the main challenges in supercomputing in the coming decade and to suggest strategic approaches for the field.

Karl Solchenbach
Director European Exascale Labs Intel
Title: The Path to Exascale

While today’s high-end computers operate in the Petascale range, future challenges like climate change, searching for new energy resources, and personalized medicine will need computers of even higher performance in the Exascale range. Building, programming and operating such exascale computers is a huge challenge since several topics need to be addressed and solved.
Energy: Based on today’s technology an Exascale computers would consume several hundred MWs. Bringing the energy (for power and cooling) down to the order of 10 MW remains a big challenge. Technologies like near threshold voltage will be necessary to minimize the energy of data processing and transfer operations.
Parallelism: An exascale computer will consist of several million cores. Only a few applications can scale to this level without significant changes. In addition to the classical strong and weak scaling schemes, applications developers and users will look into new scaling concepts, e.g. based on coupling of codes and ensemble scaling.
Resilience: The huge number of components will require special means to provide resilience, on HW and SW level.
Memory: New memory technologies will enable larger memories, leading to new IO programming models.
Intel is working on all these aspects, in Europe mainly through the European exascale labs in a co-design process.

Timothy Lanfear
Solution Architect, Nvidia
Title: Energy Efficient Computing with GPUs

The past six years have seen the use of graphical processing units (GPUs) for computation grow from being of interest to a handful of early adopters to a mainstream technology used in the world's largest supercomputers. One of the attractions of the GPU architecture is the efficiency with which it can perform computations. Energy efficiency is a key concern in the design of all modern computing systems, from the lowest power mobile devices to the largest supercomputers; it will be paramount in the push to exascale computing. We discuss innovations in processor architecture, and how the NVIDIA GPUs might evolve over the coming five years.

Lugar

Universitat Politècnica de Catalunya
Campus Nord UPC
Edificio A3, Aula Master