This year’s International Supercomputing Conference (ISC’14), which will convene in Leipzig, Germany this summer (June 22 through 26), reflects how much high performance computing has changed in the last quarter of a century. The days of HPC solely as an academic exercise for researchers and computer scientists are over. ISC’14 will be as much about commercial HPC as it is about the latest and greatest supercomputers that inhabit national labs.
That’s as it should be. Over the last three decades, we’ve seen commercial HPC in manufacturing, medicine, energy exploration, and financial services grown tremendously. In aggregate, these sectors now dominate the HPC market. As a result, commercial use now shapes a great deal of vendor behavior from HPC suppliers.
This new reality earned the new user community a separate two-day track at ISC’14, aptly named Industry Innovation through HPC. It’s designed for engineers, business types, and other HPC practitioners (or would-be practitioners) looking to use high performance computing to raise the bottom line. The track offers practical advice about the types of technologies to invest in and how they can deliver the requisite ROI.
During the industry innovation track there will be presentations on how modern clusters can be configured and used in a business environment (Evolution of Advanced Clustering and Solving Complex Problems with Affordable HPC Systems), the use of Computer Aided Engineering on HPC platforms (CAE Solutions for HPC Clusters), the challenges of mixing Windows clients with Linux servers (Stranger in a Strange Land: Running Linux Clusters in Microsoft Environments), various efforts that encourage HPC use by industry (HPC Cloud-Based Simulation Services for Mid-Caps & SMEs – First Results from the EU I4MS Initiative and HPC Impact on U.S. Industry Innovation), approaches for bringing novices into the HPC fold (Support Structures for HPC in Industry), and commercial success stories (Cloud & Big Data: Examples from Industry and Real Life Applications).
The most important element enabling the commercial spread of HPC is the reduced cost of these systems, a trend driven principally by Moore’s Law, hardware commoditization, and open source technologies. A recent development in this regard is the creation of the OpenPower Foundation, an initiative by IBM to challenge x86 dominance by opening up its Power processor to the IT community. The objective is to build a broad-based Power ecosystem that rivals that of the x86. Thus far, the Foundation has attracted an impressive array of companies, including Google, Samsung, NVIDIA, Hitachi, Mellanox, Micron, Tyan, and a handful of others.
Whether OpenPower succeeds in creating the critical mass needed to counterbalance the massive x86 ecosystem remains to be seen. To get the latest on how this is playing out in the HPC world, ISC’14 will feature a special session on the topic. Representatives from IBM, NVIDIA, and Mellanox will describe the current efforts directed at Power-based computation and outline how this dovetails into another trend sweeping across HPC: heterogeneous computing. With any luck, we’ll hear some details on what the future Power-GPU supercomputers will look like.
The commercial world aside, there will still be plenty of traditional supercomputing at the conference. A trio of presentations on exascale computing, now politely referred to as “extreme computing,” will be delivered on Tuesday, June 24. Chaired by supercomputing sage Thomas Sterling, the three talks in this session will touch on the technologies, system architectures, and programming models being considered for the next generation of machines. One of the big themes will be energy efficiency.
As befits ISC, the exascale session will draw from an international perspective: Satoshi Matsuoka, from the Tokyo Institute of Technology, will focus on power issues (Ecosystem of Extreme Computing Challenges); William Harrod, representing the US Department of Energy, will talk about advances in system software (Challenge-Driven Initiatives in Extreme-Scale Computing); and Thomas Schulthess, from CSCS in Switzerland, will describe the work being done to bring climate simulations to exascale levels (Challenges in Climate Simulations at Extreme Scale).
Energy efficiency is not just for exascalers. Power use is emerging as a limiting factor for practically every HPC datacenter. On Thursday, June 26, Natalie Bates, who chairs the Energy Efficient High Performance Computing Working Group, will direct a session on this topic (Breaking Paradigms to Meet the Power Challenges). It will feature two presentations: the first describing why BMW moved their HPC center to Iceland (hint: cheap hydro power and free air-conditioning), and the second discussing the energy savings that can be achieved with SoC technology that integrates heterogeneous cores, NICs, stacked memory, and on-chip networks.
All of that just scratches the surface of ISC’14. In addition to the sessions described here, there will be numerous Birds of a Feather Sessions, keynotes (one per day), tutorials, workshops and special panels, not to mention the three-day exhibition, where vendors will display their latest HPC wares. And as always, there will be the semiannual refresh of the TOP500, the acclaimed (or for some, reviled) list of the fastest computers in the world.
Whether you run a $100 million supercomputer at a national lab or a 4-node cluster beside your desk, or you just want to learn about the latest and greatest in HPC, there will be plenty of ways to fill your time at ISC’14. Check out the online program for the complete agenda.