Oklahoma Supercomputing Symposium
Permanent URI for this collection
Browse
Recent Submissions
Item Open Access Building an Exotic HPC Ecosystem at The University of Tulsa(2015-09-23) Hale, John; Hawrylak, Peter; Andrew, KongsThis talk covers the in-progress journey of the Tandy School of Computer Science at The University of Tulsa to build a unique high performance computing (HPC) ecosystem for researchers and students. The presenters motivate and describe the launch of TU's initial HPC point of presence — a traditional CPU cluster — along with lessons learned from that process. They also discuss ongoing work to stand up two distinct heterogeneous compute node clusters and the challenging research problems they will be used to address. Objectives and developments in leveraging these HPC resources in the classroom will be presented. In addition to passing along some wisdom picked up along the way, the presenters will reveal their plans for the future of TU's evolving HPC ecosystem.Item Open Access A Data Communication Reliability and Trustability Study for Cluster Computing(2015-09-23) Colmenares, EduardoAbstract In High Performance Computing (HPC), most of the problems under study will be either embarrassingly parallel or data dependent. Beyond the nature of the problem, scientists will be interested in either one or two additional characteristics. The first, performance, focuses in achieving an accurate solution in a fraction of the time of a sequential approach. The second is consecutive, accurate and steady time readings. In their quest for performance, some scientists forget not only that the chosen tool, in many cases a distributed-memory system, is a multi-user system, but also that its components are interconnected through a high-speed communications network to facilitate the interaction among processors. In this talk, we show why a cluster characterization is relevant, particularly for scientific kernels where multiple accurate and consecutive time readings are necessary to statistically validate a behavior. We present the characterization of two clusters by using two variants of the ping pong test. One of the clusters is a multi-user research oriented cluster, while the second is a one-user cluster with older technology.Item Open Access Software Defined Networking – That's the answer, What's the question?(2015-09-23) DeBacker, DanOh no, not another SDN presentation talking about a bunch of new techie acronyms that mean nothing to me. Well, there will be some of that here, however in this presentation you'll also get a perspective on SDN in regard to the reality of its use. There is no doubt that SDN will touch every network in some shape, form or fashion in the future. How and to what extent will vary greatly. This presentation will focus on the technologies of SDN and real life use case implementations to solve real life issues.Item Open Access Computing Hydrogen Ion Survival Probability: Academy Student, Graduate Student, and Faculty Experiences(2015-09-23) Monismith, David; Zhang, Yixiao (Icy); Shaw, John; Chakraborty, HimadriThis presentation covers the experiences of a Missouri Academy student, a Graduate Directed Project team, and Computer Science and Physics Faculty at Northwest Missouri State University in data management, computational science and physics while simulating firing a Hydrogen Ion at a metal surface. Faculty involved in the project, Drs. Chakraborty, Monismith, and Shaw, were awarded XSEDE startup and XRAC allocations to perform over 20,000 2D simulations of firing a hydrogen ion at various metallic surfaces at a scale of hundreths of atomic units. Simulations in this project allowed for variations in the trajectory model used, distance of closest approach, normal velocity, parallel velocity, height of the potentials, width of each potential, and distance between adjacent steps. Academy student experiences included learning about directive based parallelism and updating a Fortran IV/77 code to Fortran 90 and to include OpenMP parallelism. Graduate students involved in a graduate directed project developed two codes as part of a data management plan for the project. The first was to upload simulation results from the TACC Stampede supercomputer to a server at Northwest Missouri State University to retain results in a MySQL database. The second was to retrieve data from this MySQL database and present it in a graphical format using a Java Swing GUI tool that produced graphical reports using the JasperReports API. Faculty have performed significant optimizations to the code to allow for single parameter set executions that make use of all compute resources on a Stampede node - asynchronous OpenMP/Xeon Phi OpenMP with 16 and 240 cores, respectively. So far results in this project have been produced for two metals and Drs. Chakraborty and Shaw have over 30 graphs on which they are performing analysis. Dr. Monismith is currently performing optimizations on a 3D version of this code on PSC Greenfield.Item Open Access On the Centrality of HPC for taking NGS to the next frontier: Clinical application at scale(2015-09-23) Wheat, Stephen; Martinez-Canales, MonicaNext Generation Sequencing (NGS), the basis for volume sequencing enablement, has been around for several years. The volume of sequencers deployed per year remains on an exponential growth path. Nevertheless, the vision of sequencing-enabled personalized medicine has come to fruition for relatively few people. The community consensus is that bringing this to large populations remains 5-7 years out. Nevertheless, some projects are underway to path-find means to accelerate this. In this talk, we will review the solution architecture that will enable this from a technology perspective. Furthermore, we will review the efforts of the Intel/HP HPC Alliance with respect to driving these solutions into actual implementation. While the solutions architecture will be focused on the NGS work flow, the elements of the architecture are pertinent to other HPC work flows.Item Open Access OSCER State of the Center Address 2015(2015-09-23) Neeman, HenryThe OU Supercomputing Center for Education & Research (OSCER) celebrated its 14th anniversary on August 31 2015. In this report, we examine what OSCER is, what OSCER does, what OSCER has accomplished in its 13 years, and where OSCER is going.Item Open Access Exploring Adverse Drug Effect Data with Apache Spark, Hadoop, and Docker(2015-09-23) Davis, NicholasAdverse drug reactions (ADRs), a subset of the broader adverse events (AEs), have been shown in several studies to have a considerable burden on healthcare costs and patient outcomes. ADRs account for a significant increase in patient morbidity, mortality, and additional healthcare costs. In this presentation, we explore ADRs and AEs from the U.S. Food and Drug Administration's Adverse Event Reporting System (FAERS) data set. Using big data analysis tools from the Hadoop ecosystem, including Apache Spark, we analyze the FAERS data and discuss interesting trends and observations in the 10+ year historical data set.Item Open Access Parallel programming in the classroom : analysis of genome data(2015-09-23) Frinkle, Karl; Morris, MikeOver the course of a semester, students enrolled in an HPC seminar class created a suite of human genome analysis tools on the Beowulf clusters that they and other students built. The analysis tools were written with C and MPI and subsequently interfaced with a LAMP (Linux, Apache, MySQL, and PHP) website through the use of scripts. The output was visualized with the help of Google Charts. We will discuss the technical details of this project and demonstrate how these tools can be used to analyze multiple human genomes simultaneously. Presented at: 2015 Oklahoma Supercomputing Symposium Wednesday September 23 2015, 8:00am-5:00pm Thurman J. White Forum Building, 1704 Asp Ave., Norman OK 73072Item Open Access Big Storage, Little Budget(2015-09-23) Hutson, Kyle; Tygart, Adam; Andresen, DanKansas State University's HPC cluster was running out of storage space last year. Vendors of traditional HPC storage solutions were either too expensive to be feasible or too little capacity to be of long-term use. The system that ended up providing the best storage capacity for the available budget was Ceph, an open-source project that provides storage striped across many commodity servers. This session is a case study of the pros and cons of our implementation of a 1.5 PB Ceph-based storage cluster, discussing the history of network-based filesystems, including why our previous Gluster-based was no longer suitable. Questions and discussion are encouraged.Item Open Access Enabling Computational Research and Education at the University of Central Oklahoma(2014-09-24) Lemley, EvanLessons learned creating the Center for Research and Education in Interdisciplinary Computation (CREIC) - Persistence, Persistence, Persistence - Internal Help/Support - External Help/Support (OneOCII) - Pay attention to reviews - If it’s clear someone did not read what you wrote -- figure out why... for some - Line-by-line read throughs - Visit a Program Officer after you get turned down (but quite a bit before next proposal is due) - Leverage existing funding, activity, momentum on your campus Overview - Background - College of Mathematics and Science (CMS) - CREIC - MRIItem Open Access Advancing Research Computing on Campus and Beyond(2014-09-24) Brunson, DanaThe demand for computational resources, support, and expertise among the academic community continues to grow with each passing year and can quickly surpass an institution's capacity to meet researchers' needs. The return on investment in these resources can lag behind the initial investment causing challenges in developing an institutional commitment for funding centralized highly available cyberinfrastructure. In this session, we will discuss the approach taken at OSU to develop these resources utilizing external funding, cooperation and collaboration with others, and attention to specific needs of the campus community.Item Open Access Getting HPC into Regional University Curricula with Few Resources(2014-09-24) Frinkle, Karl; Morris, MikeWe showcase our successful project of injecting high performance computing (HPC) into the traditional computer science curriculum at Southeastern Oklahoma State University. We incorporated a three-semester sequence of parallel programming courses, with the third course focusing on a research-level mathematical project that was executed on OU's supercomputer. Emphasis was placed on utilizing Open MPI and CUDA libraries along with parallel algorithm analysis and file I/O optimization. We recruited students with varied CS backgrounds for the program, some with only CS-1 in their portfolios. These courses are currently being followed with a more hardware-oriented course this fall, which will be profiled in the talk. We will discuss our approach to making this a successful sequence of courses and divulge many of the tips and tricks we have learned from our efforts.Item Open Access Synergistic Opportunities for Computational Resource Providers and Local IT(2014-09-24) Gray, ZaneTalk Abstract How do high-speed data paths benefit research computing, and should you care? This talk explores the means of engaging local Information Technology resources to build two-way, collaborative partnerships that enable the achievement of research goals. We will discuss examples of how this relationship benefits local IT through exposure to cutting edge technologies prior to introduction into the campus or enterprise environment — a selling point in an affiliation that has traditionally been more "take" than "give." We will also discuss when circuit services, Science DMZ's, and Research and Education Networks are appropriate enough to warrant engagement with local or regional IT resources.Item Open Access Interstate Data Moving and the Last Block Problem: Lessons Learned in CAPS Spring Experiment 2014(2014-09-24) Brewster, KeithTalk Abstract In Spring of 2014, for the first time, CAPS endeavored to send complete high resolution 4D numerical weather prediction forecast files from the National Institute for Computational Sciences (NICS) in Oak Ridge TN to forecasters in the Hazardous Weather Testbed in the National Weather Center at the University of Oklahoma. Having the complete files would allow the use of 3D visualization tools and exploration of the forecast output not previously possible. As weather forecasts, the data were very perishable and prompt reliable throughput was essential. The issues around moving very large datasets in real time halfway across the country are explored from the end-user perspective, with some lessons learned and some future solutions presented.Item Open Access Rapid Calculation of Medication Adherence Using Parallel Computing with R and Python(2014-09-24) Davis, Nicholas____R and Python (with Pandas) are excellent languages for data analysis ____Parallelizing code is often trivial, with some caveats ____Faster runtimes lead to richer exploration of the dataItem Open Access OSCER State of the Center Address 2002(2002-09-13) Neeman, HenryOSCER State of the Center Address OU Supercomputing Symposium 2002 Thursday September 12 - Friday September 13 2002Item Open Access Birds of a Feather Session: Production Cluster Deployment - Then, Now, and Future Trends(2010-10-06) Brunson, Dana; Pummill, JeffBoF Abstract: This BoF is intended as an introduction to the many components that comprise clustered HPC systems; past, present, and a brief look at future trends. The presentation and accompanying discussion will address topics such as: what to look for in a vendor, how to choose hardware type(s), power and cooling issues, cluster software stacks, administrative tips and tricks, user support advice including scheduling policies and file management, and hopefully a lively debate at the end. This BoF is not intended to define what should and should not be deployed; rather, we will present a number of factors and considerations involved in deploying a successful cluster.Item Open Access OSCER State of the Center Address 2003(2003-09-25) Neeman, HenryTalk Abstract OSCER State of the Center Address 2003 The OU Supercomputing Center for Education & Research has undergone tremendous growth in the past year, not only in size but also in scope and ambition. In this talk, we'll examine where we are, how we got here, and where we want to go.Item Open Access OSCER State of the Center Address 2002(2002-09-12) Neeman, HenryThe State of the OU Supercomputing Center for Education & Research (OSCER)Item Open Access OSCER State of the Center Address 2012(2012-10-03) Neeman, HenryTalk Abstract The OU Supercomputing Center for Education & Research (OSCER) celebrates its 11th anniversary on August 31 2012. In this report, we examine what OSCER is, what OSCER does, what OSCER has accomplished in its 11 years, and where OSCER is going.