Additional information can be found at:
https://support.opensciencegrid.org/support/solutions/articles/12000025149-osg-managed-services
Dear reader, how should you disseminate your software? If you want your recipe to come out just right, we encourage you to put it in a container. One such container, Singularity, is the first of its kind to be securely deployed internationally on more than 40 shared cluster resources. Its registry, Singularity Hub, further supports reproducible science by building and making containers accessible to any user of the software. In this talk, Vanessa will review the primary use cases for both Singularity and Singularity Hub, and how both have been designed to support modern, common workflows. (Greg will participate remotely.) She will discuss current and future challenges for building, capturing metadata for, and organizing the exploding landscape of containers, and present novel work for assessing reproducibility of such containers. Containers are changing scientific computing, and this is something to be excited about.
Attracting, retaining and developing female talent across the world is not only essential to an organization performance-- it's a business imperative. Learn how to be a leader of positive change for women in HPC. Learn how to create the social infrastructure to give women and other underrepresented groups access to the education, resources and opportunities they need to fully reach their potential. Also hear about how important it is to be focused on the pipeline of girls staying in science, technology, engineering and math in the learning paths. Come to hear about what IBM Canada and XSEDE are doing to help drive engagement of girls in STEM and be prepared to have an open discussion about what still needs to be done to increase the numbers of girls pursuing STEM education and careers.
PresentersScience is being enabled by Supercomputing, whether it’s climate science, combustion science, or understanding the fundamentals of how the human body works. What’s exciting is that the same technology enabling this powerful science is also enabling the revolution in deep learning, and it’s being enabled by GPUs. This session will explain how the rapid advancement of deep learning for artificial intelligence has created an enormous demand for computational resources outside the traditional supercomputing domains. NVIDIA is uniquely suited to address these evolving needs with accelerated computing and will present recent GPU hardware and software advances and how they address computational needs in both AI and HPC. There will also be an update on the Deep Learning Institute (DLI) Teaching Kit, which NVIDIA has partnered with Professor Yann LeCun of New York University and Facebook AI Research. The kit covers the academic theory and application of deep learning on GPUs using the PyTorch and Torch frameworks and includes detailed lecture slides, hands-on labs/source code solutions, quiz/exam problem sets, and free access to online deep learning labs using GPUs in the cloud.
Participants will divide into 3 groups selected by campus champions - BoF attendees to select the topic group of their choice. Each group will discuss their topic for ~ 45 minutes; at the culmination of the individual group discussions, each group lead will be allotted 5 minutes for both a lightning summary talk on their topic.
1. Designing and sustaining a financial model for CI to enable centralized and shared resources institutionally. As the number of campuses involved in research computation continues to grow, a primary concern is that of sustaining the CI program financially over a broad course of time.
3. Storage Models for Research Computing
The proliferation of data continues to grow rapidly from the availability of large research repositories and growth of local instrumentation with high resolution. This increase in the volume of data been generated, has created a lot of pressure for Research Computing organization to create a sustainable model for storage that is flexible, scalable, and low cost. A well organized plan (or lack thereof) can have a profound impact on the local research community.
In this presentation we will discuss best practices and methodology for HPC software engineering. We will provide illustrations of how the Allinea debugging and performance analysis tools can be used to ensure that you obtain optimal performance from your codes and that your codes run correctly.
In this talk Roger Goff, DDN’s Senior Technologist for Academic Research, will present 3 customer case studies highlighting common IO problems in shared research environments and recent examples of how advanced computing centers are solving them with innovative shared infrastructure.
Examples will highlight problems and practical solutions for fast data tier, active archive, deep archive and multi-site collaboration and data protection. Examples covered will include technologies specifics on NVMe SSD, Flash Caching, Automated Tiering, Private/Hybrid Cloud Infrastructure, Active Archive, multi-site shared namespace and production scale neural networks.
The topic is broken into three components centered around efficiency in: data lifecycle, job scheduling and operational management. Through a series of tools and processes, DST will show the research community how to more effectively utilize the HPC environment resources.
The genomic revolution has led to a significant drop in the cost of sequencing of entire organisms including humans. This has led to a better understanding of the basic building blocks of life, diseases and increased medical knowledge. In turn, this has resulted in a plethora of new diagnostics tests and drugs. In particular advances in medical technologies such as medical imaging, molecular modeling and therapeutic devices have led to an exponential growing data deluge in the life sciences. As a result, organizations are spending millions of dollars on IT infrastructure and IT support both on-premises and in the Cloud to store, analyze and access this data. Many of these organizations lack a dedicated research IT infrastructure and scientific IT support, their existing enterprise IT is ill prepared to deal with scientific workflows resulting in a large gap between science and IT. This is because IT has traditionally been the group that manages only the business and enterprise systems for an organization such as desktop support, email, web-services, HR, or databases. The consequence of this lack of a dedicated research IT within an organization is impacting the research of scientists and labs at different levels, from day to day IT related activities, to advanced scientific computing needs for large-scale analytics, to collaboration and data exchange with peers both within and outside of the organization.
An overview of the architecture and scientific uses of the HPC “Bridges” system which resulted from a partnership between Pittsburgh Supercomputing Center (PSC) and Hewlett-Packard Enterprise (HPE). Bridges is a uniquely capable resource for empowering new research communities and bringing together HPC and Big Data.
Domain science experts are commonly limited by computational efficiency of their code and hardware resources available for execution of desired simulations. Here, we detail a collaboration between domain scientists focused on simulating an ensemble of climate and human management decisions to drive environmental (e.g., water quality) and economic (e.g., crop yield) outcomes. Briefly, the domain scientists developed a message passing interface to execute the formerly serial code across a number of processors, anticipating significant performance improvement by moving to a cluster computing environment from their desktop machines. The code is both too complex to efficiently re-code from scratch and has a shared codebase that must continue to function on desktop machines as well as the parallel implementation. However, inefficiencies in the code caused the LUSTRE filesystem to bottleneck performance for all users. The domain scientists collaborated with Indiana University’s Science Applications and Performance Tuning and High Performance File System teams to address the unforeseen performance limitations. The non-linear process of testing software advances and hardware performance is a model of the failures and successes that can be anticipated in similar applications. Ultimately, through a series of iterative software and hardware advances the team worked collaboratively to increase performance of the code, cluster, and file system to enable more than 100-fold increases in performance. As a result, the domain science is able to assess ensembles of climate and human forcing on the model, and sensitivities of ecologically and economically important outcomes of intensively managed agricultural landscapes.