Search PNNL

• PNNL Home
• About
• Research
• Publications
• Jobs
• Newsroom
• Contacts

SC11 PNNL Leadership in Technical Sessions

Paper Sessions

An Early Performance Analysis of POWER7-IH HPC Systems

Wednesday, November 16, 2:00 p.m. – 2:30 p.m.
TCC 303

In this work we present a performance evaluation of the POWER7-IH processor and of integrated systems built from it. We describe the architecture of P7-IH with an emphasis on those characteristics that have a direct impact on the performance for large-scale HPC systems and applications. An important area of emphasis is the memory and communication subsystems and their impact on achievable application performance. The results from a set of micro-benchmarks are presented that include memory, communication and OS-noise characteristics. In addition the results from several production level applications are analyzed and their performance linked to the results of the micro-benchmarks through the use of accurate performance models. The models will also be employed in exploring the achievable performance of these applications on much larger systems.

Scalable Implementations of Accurate Excited-state Coupled Cluster Theories: Application of High-level Methods to Porphyrin-based Systems

Thursday, November 17, 3:30 p.m. – 4:00 p.m.
TCC 304

The development of reliable tools for excited-state simulations is important for understanding complex processes in the broad class of light harvesting systems and optoelectronic devices. Over last years we have been developing equation of motion coupled cluster(EOMCC) methods capable of tackling these problems. In this paper we discuss parallel performance of EOMCC codes which provide accurate description of excited-state correlation effects. Two aspects are discussed in detail:(1) a new algorithm for the iterative EOMCC methods based on improved parallel task scheduling algorithms, and (2) parallel algorithms for the non-iterative methods describing the effect of triply excited configurations. We demonstrate that the most computationally intensive non-iterative part can take advantage of 210,000 cores of the Cray XT5 system at the Oak Ridge Leadership Computing Facility(OLCF), achieving over 80% parallel efficiency. In particular, we demonstrate importance of computationally demanding non-iterative many-body methods in matching experimental level of accuracy for several porphyrin-based systems.

---

Panel Sessions

Holistic Co-Design Approach using Application Performance Modeling and Simulation for System Design, Evaluation, and Optimization

Tuesday, November 15, 3:30 p.m. – 5:00 p.m.
TCC 101

Future Petascale++ systems will use "co-design" as a key method to acceleration deployment for both systems and applications. Blue Waters is the first and largest HPC system project that extensively used application and system co-design principles in both development and deployment phases. A critical part of the Blue Waters co-design effort is an unprecedented level of holistic modeling and simulation approach to application re-engineering and system design. This panel will bring together five independent yet synergistic modeling, simulation and performance prediction approaches that engaged with the multiple Blue Waters science teams and system developers. The multi-faceted effort simultaneously helps analyze and optimize both the Blue Waters system and the applications that will use it. This panel presents the integrated modeling and simulation efforts, the impacts these efforts are having on both the system design and application implementations and lessons learned that can be applied to trans-petascale and exascale development efforts.

Scientific Data on the Path to Exascale: Lessons, Insights and Predictions from 10+ years on the front lines

Wednesday, November 16, 3:30 p.m. – 5:00 p.m.
TCC 101

This panel will present an overview of how data-centric research has made a difference to scientific applications within the SciDAC program, what advances have enabled data management and analysis to keep up with the increasing power of the supercomputers driving the simulations, and where the discipline is headed as we continue to move towards exascale computing. During their 20 minute presentations, the panelists will be encouraged to take strong positions while answering the following questions regarding data-centric research: - Give at least 2 examples of where data intensive research has enabled new scientific discoveries - What advancement has been the most disruptive to science over the past decade - What advancement has had the most impact on science over the past decade - What advancement has been most over-hyped within the past decade - What has not worked as well as you had expected - What advancement do you see as being the most important in the coming 5 years.

---

Workshops

2nd International Workshop on Modeling, Benchmarking and Simulation of High Performance Computing Systems

Sunday, November 13, 9:00 a.m. – 5:30 p.m.
Leonesa II

This workshop is concerned with the comparison of HPC systems through performance modeling, benchmarking or through the use of tools such as simulators. We are particularly interested in the ability to measure and make tradeoffs in software/hardware co-design to improve sustained application performance. We are also concerned with the assessment of future systems to ensure continued application scalability through peta- and exa-scale systems. The aim of this workshop is to bring together researchers, from industry and academia, concerned with the qualitative and quantitative evaluation and modeling of HPC systems. Authors are invited to submit novel research in all areas of performance modeling, benchmarking and simulation, and we welcome research that brings together current theory and practice. We recognize that the coverage of the term 'performance' has broadened to include power consumption and reliability, and that performance modeling is practiced through analytical methods and approaches based on software tools and simulators.

Climate Knowledge Discovery Workshop

Sunday, November 13, 9:00 a.m. – 5:30 p.m.
Discovery B

As we enter the age of data intensive science, knowledge discovery in simulation based science rests upon analyzing massive amounts of data. In climate science, model-generated and observational data represent one of the largest repositories of scientific data in any discipline. Geoscientists gather data faster than they can be interpreted. They possess powerful tools for stewardship and visualization, but not for data intensive analytics to understand causal relationships among simulated events. Such tools will provide unique insights into challenging features of the earth system, including anomalies, nonlinear dynamics and chaos with the potential to play a significant role in future IPCC assessments. The breakthroughs needed to address these challenges will come from collaborative efforts involving end-user scientists, computer and computational scientists in diverse disciplines to discuss the design and development of methods and tools for knowledge discovery in climate science. See presentations from the first workshop.

High Performance Computing, Networking and Analytics for the Power Grid

Sunday, November 13, 9:00 a.m. – 5:30 p.m.
Discovery A

The workshop intends to promote the use of high performance computing and networking for power grid applications. Technological and policy changes make this an urgent priority. Sensor deployments on the grid are expected to increase geometrically in the immediate future, while the demand for clean energy generation is driving the use of non-dispatchable power sources such as solar and wind. New demands are being placed on the power infrastructure due to the introduction of plug-in vehicles. These trends reinforce the need for higher fidelity simulation of power grids, and higher frequency measurement of their state. Traditional grid simulation and monitoring tools cannot handle the increased amounts of sensor data or computation imposed by these trends. The use of high performance computing and networking technologies is of paramount importance for the future power grid, particularly for its stable operation in the presence of intermittent generation and increased demands placed on its infrastructure.

IA³ - Workshop on Irregular Applications: Architectures & Algorithms

Sunday, November 13, 9:00 a.m. – 5:30 p.m.
Favorita

Many data intensive scientific applications are by nature irregular. They may present irregular data structures, control flow or communication. Current supercomputing systems are organized around components optimized for data locality and regular computation. Developing irregular applications on them demands a substantial effort, and often leads to poor performance. However, solving these applications efficiently will be a key requirement for future systems. The solutions needed to address their challenges can only come by considering the problem from all the points of view, from micro to system-architectures, from compilers to languages, from libraries to runtimes, up to rethinking how algorithms operate. Only collaborative efforts among researchers with different profiles, including end users, domain experts, and computer scientists, could lead to significant breakthroughs. This workshop aims at bringing together scientists with all these different backgrounds to discuss, define and design methods and technologies for efficiently supporting irregular applications on current and future machines. See the Center for Adaptive Supercomputing - Multithreaded Architectures (CASS-MT) website for more information.

Workshop on High Performance Computing Meets Databases (HPCDB)

Friday, November 18, 9:00 a.m. – 3:00 p.m.
TCC 303

Current interest from the HPC community in data-intensive computing, simpler programming models, in situ analytics, ad hoc data analysis, and interactive control of long computations offers new opportunities for engagement between the database and HPC communities. The relational model for databases revolutionized business data management, leading to a $30B market and a vibrant startup community. The key idea is to "push computation to the data," insulating applications from details of data representation and system architecture while affording runtime optimizations unavailable to compile-time techniques. The techniques of this approach - a rigorous data model, declarative query languages, cost-based optimization, logical and physical data independence - have been applied to new data types (streams, graphs, arrays), new applications (imaging, genomics, GIS, finance) and new platforms (embedded, GPGPUs, cloud). But this approach has only been minimally explored in HPC contexts. In this workshop, we invite position papers that explore this space.

---

Birds-of-a-Feather Session

Semantic Graph Database Processing

Tuesday, November 15, 12:15 p.m. – 1:15 p.m.
TBD

As the size of data and complexity of the relationships increases, relational databases and SQL queries are failing to fulfill query requirements. The more general representation of data in RDF triples (semantic graphs), as opposed to relational tables, is a natural evolution for data/information storage. In the midst of this migration of data storage and retrieval strategies, the exploding size of data is hindering existing SPARQL query engines from performing the types of queries that this semantic graph technology promises. This BOF aims to cultivate a community of HPC-based Semantic Database (SDB) stakeholders. Attendees will find out what others are working on in this nascent area that blends HPC with SDBs. They will have an opportunity to participate in the formation of a new direction with the potential for high impact for many years. Potential researchers, developers, and users are all welcome.

Cray XMT

Wednesday, November 16, 12:15 p.m. – 1:15 p.m.
TBD

The Cray XMT is a shared-memory, multithreaded computer platform well suited for applications with little or no spatial and temporal locality. The system's large shared-memory, latency tolerant processors, and fine-grained hardware synchronization enable highly irregular memory access patterns and algorithms to scale. Researchers at national laboratories, universities, and industry are using the machine to develop large-scale applications in bioinformatics, network analysis, semantic analysis, natural language understanding, and knowledge discovery. Moreover, numerous semantic technologies are being developed for the system.

This year Cray has delivered the first Cray XMT/2 systems. This Birds-of-a-Feature will bring together the massive multithreading community to discuss results, current status, and future machine designs driven by application requirements. The meeting will feature talks on use cases, applications, performance, programming tools, and expectations of next-generation multithreaded systems.

Energy Efficiency at Extreme Scales: Tools and Challenges

Wednesday, November 16, 5:30 p.m. – 7:00 p.m.
TCC LL3

Power consumption will become a major concern for future generation supercomputers. Current systems consume more than a Megawatt per Petaflop. Exascale levels of computation, a 100x improvement in performance from today, will be significantly constrained if power requirements were to similarly scale. The optimization of power and energy at all levels, from application to system software and to hardware, is required. This BOF will discuss state-of-the-art tools and techniques for observing and optimizing power consumption. These are often limited by hardware capabilities and in their ability to associate measurements with software. Additionally, control mechanisms such as Dynamic-Voltage-Frequency-Scaling are often limited. The challenges ahead are many-fold. Increasing parallelism, memory systems, interconnection networks, storage and uncertainties in programming models all add to the complexities. More rapid realization of energy savings will require significant increases in measurement resolution and optimization techniques. The interplay between performance, power, and reliability also leads to complex trade-offs.

New Developments in the Global Arrays Programming Model

Thursday, November 17, 5:30 p.m. – 7:00 p.m.
WSCC 2A/2B

The purpose of this BOF is to obtain input from the Global Array user community on proposed development of the GA toolkit. The main focus will be on extending the GA programming model to post-petascale architectures but other topics, including the addition of desirable features relevant for programming on existing platforms will also be entertained. The discussion will focus on addressing bottlenecks to scalability, including fault tolerance, data consistency in one-sided communication models, and flexible distributed data structures. The session format will be geared towards informal discussion and will invite extensive input from session attendees. The session leaders will provide a brief overview of the GA programming model and then discuss issues associated with extending the GA programming model to the next generation of computers and current plans to deal with them. The session will then be opened up for discussion and comments by session participants.

Security & Privacy | Site Index | Contact PNNL