Birds of a feather sessions

Three parallel sessions are scheduled for 7:00PM - 9:00 PM on Dec 19.
6:00 PM to 7:00 PM: Invite-only dinner provided by BOF organizers. Contact organizers for dinner invites.

Please register for individual BoF sessions. More information on registration is available below.

Nvidia

Location: (Majectic II) (Registration & full details)

Topic	Presenter
Kepler K20 Launch: World's fastest Accelerators	Sanjiv Satoor
CUDA 5.0 feature update: Making Programming With World's Most Pervasive Parallel Computing Platform Even Easier	Pradeep Gupta
OpenACC : Directive-based GPU programming overview	Pradeep Gupta
Application Performance on GPUs	Ananda Bhattacharya

Mellanox

Location: (Royale I) (Registration & full details)
Designing InfiniBand fabrics for next-Gen HPC and Cloud
Through this BoF session we would like to bring together people interested in designing Infiniband HPC clusters and deploying next-Gen Cloud infrastructure. We will focus on Infiniband topologies, fabric configuration, management and troubleshooting. Discuss various high-performance storage options for HPC, block storage and virtualization technologies for Cloud deployments. Attendees can also deliberate FDR/QDR, network oversubscription and how to effectively design cluster fabric suitable for various workloads.

Fujitsu

Location: (Majestic I) (Registration & full details)
Introduction to 'challenges to achieve real Petascale computing in Japan' from various aspects.
Talk 1

Title : Recent Japanese challenge for High Performance Computing - K computer and future
Speaker : Dr. Motoi Okuda, Fujitsu Ltd
Abstract : Japanese government are strategically expanding investment to HPC. My talk aims to cover the outline of the ongoing Japanese HPC project, the K computer project and HPCI : HPC application project. Japanese government has just begun their new challenge to the future supercomputer system. I will also refer to the outline of Japanese activities, based on lessons we learned from K Computer project Duration: (40min. with 5 min Q&A, including K computer making video)

Talk 2

Title : Introduction of the K computer and its successor PRIMEHPC FX10
Speaker : Takayuki Hoshiya, Fujitsu Ltd
Abstract : K computer is the first 10PFLOPS system and PRIMEHPC FX10 is its successor developed by RIKEN and Fujitsu. In this talk, I introduce key technologies for petascale computer and how they are implemented in K computer and FX10. SPARC64 VIIfx, IXfx processor had extended SPARC architecture called HPC-ACE. Its concepts and performance in HPC application would be presented. Also petascale system configurations, system software for using huge system effectively, and technologies to achieve lowpower consumption and high reliability would be covered. Duration: (40min. with 5 min Q&A )

Talk 3

Title : HPC Research and Open Innovation in Fujitsu Laboratories of Europe
Speaker : Dr Nick Wilson (Fujitsu Laboratories of Europe)
Abstract : High-performance computing (HPC) is a vital tool in ensuring a safe and prosperous future society. Fujitsu Laboratories of Europe (FLE) is using open innovation to develop software technologies that increase the effectiveness of HPC applications on current and future supercomputers. Examples include the Open Petascale Libraries project, which promotes the development of numerical libraries for massively parallel computers based on multi-core nodes; work with The Australian National University targeted at fault-resilient software; and the HPC Wales project, which seeks to accelerate industrial competitiveness through use of HPC in small and medium enterprises. Examples will also be given of two recent HPC application projects in FLE: in silico drug testing, which will contribute to cost-effective healthcare, and tsunami modelling to mitigate the effects of future natural disasters. Duration: (40min. with 5 min Q&A )