Extended Collaborative Support program provides expert assistance in a wide range of
cyberinfastructure technologies. Any user may request this assistance through the XSEDE allocation process.
The primary goal of this monthly symposium is to allow the over 70 staff members working in ECSS to exchange information about successful techniques used to address challenging science problems. Tutorials on new technologies may also be featured. Two 30-minute, technically-focused talks will be presented each month and will include a brief question and answer period. This series is open to all.
These sessions will be recorded. For this large webinar, only the presenters and host will be broadcasting audio. Attendees may submit questions to the presenters through a moderator by sending a chat message.
February 21, 2017
Julia on HPC Platform
Presenter(s): Dong Ju Choi (UCSD)
Principal Investigator(s): Christopher Rackauckas (University of California, Irvine)
Julia is a high level programming languge providing both MATLAB like easiness and parallel computing performance. We began to learn the language to assist Chris Rackauckas at University of California Irvine for the development of a Julia differential equation package on SDSC Comet system.
This presentation will introduce the basic Julia usage on HPC platform specially with Comet system during the support.
Presenter(s): Anirban Jana (PSC)
Principal Investigator(s): Metin Sitti (CMU)
In this talk, I will present my ECSS project with Prof Metin Sitti from CMU, where I helped develop an efficient simulation capability to perform simulations of microrobots in liquid enviroments at low Reynolds numbers (in the Stokes flow regime). The simulations are based on the boundary element method for Stokes flows. The ECSS project activities included selection of a suitable software stack, workflow development from preprocessing to simulation to analysis, and parallelization of an existing open source Stokes flow BEM code . This enabled simulation time to be reduced from days to minutes, opening up the possibility of comprehensive design space explorations, or more highly resolved simulations, or simulations of much more complex systems such as swarms of microrobots, in the future.
January 17, 2017
Toward Flood Inundation Mapping at the Continental Scale: the Cyberinfrastructure Approach
Presenter(s): Yan Liu (CyberGIS Center and NCSA, UIUC)
Principal Investigator(s): David Maidment (UT Austin) David Tarboton (Utah State University)
Project 1: National Flood Interoperability Experiment (NFIE) As high-resolution national terrain and water data become increasingly available, hydrology researchers have the opportunities to conduct hydrological study directly at continental scale for the conterminous U.S. (CONUS). To enable continental hydrology research, integrated cyberinfrastructure power plays a critical role in coupling interdisciplinary data, software, and multiple types of computational platforms and making them available for the research community to perform methodological experiments on big data and complex computations that they could not handle in lab environment. One leading effort in this direction is the NSF National Flood Interoperability Experiment (NFIE). NFIE aims to combine high-resolution terrain (10m and finer) and water (National Hydrography Dataset) data with NOAA National Water Model (NWM) forecast for real-time national inundation mapping and forecast.
Project 2: Terrain analysis using digital elevation models (TauDEM) Two ECSS projects help achieve NFIE goals on XSEDE resources at NCSA (on ROGER supercomputer, one of the 3rd-tier resources) and TACC (on Stampede). The NFIE ECSS project develops computational solutions to data integration and processing, methodology development, workflow construction, and experiment computation. The TauDEM ECSS project accelerates a core software piece in NFIE workflow, i.e., TauDEM, to conduct high-performance and scalable hydrological information analysis on large geospatial raster data. In this talk, I will introduce the data and computational challenges in these two projects and review the solutions developed to achieve major milestones. Results and their impact in the research community will be presented. I will share our computational experience on the highly-coupled HPC and cloud computing platform on ROGER and discuss the advantages of hybrid supercomputing architecture in building end-to-end solutions for research groups and science gateways.
December 20, 2016
Comet Virtual Cluster 'User' Experience
Presenter(s): Trevor Cooper (SDSC) Fugang Wang (IU)
Comet is an XSEDE HPC resource hosted and operated at SDSC and supported by systems staff at SDSC and user support staff at IU.
This demonstration explores the Virtual Cluster (VC) capability of Comet, a unique feature that provides projects with the ability to fully define their own software environment with a set of dynamically allocated virtual machines.
We will begin the demonstration with an overview on the design and architecture of the virtual cluster capability, and how it compares to other virtualized and cloud services. The high performance of the virtualized clusters combining the full AVX2 feature set of the Haswell processors and the InfiniBand HCAs using SR-IOV for MPI will be discussed.
We will then follow with a demonstration on how to build, configure, and manage virtual clusters using the Cloudmesh client, a tool to easily interface with multiple clouds from the command line and a command shell following a guide originally create for the XSEDE hands-on tutorial. Time-permitting we will demonstrate multiple pre-configured virtual clusters and provide examples of possible administrative workflows available in this unique environment.
Finally we will provide references for further information on obtaining an allocation for Comet virtual clusters.
The Science Gateways Community Institute
Presenter(s): Nancy Wilkins-Diehr (SDSC)
Science gateways are a fundamental part of today's research landscape. Beginning in 2013, more users accessed XSEDE resources via gateways than they did from the command line. However, despite the presence of gateways for many years, development of these environments is often done with an ad hoc process, limiting success, resource efficiency, and long-term impact. Developers are often unaware that others have solved similar challenges before, and they do not know where to turn for advice or expertise. Without knowledge of what's possible, projects waste money and time implementing the most basic functions rather than the value-added features for their unique audience. Critically also many gateway efforts fail. Some fail early by not understanding how to build communities of users; others fail later by not developing plans for sustainability.
The Science Gateways Community Institute (SGCI, http://www.sciencegateways.org) is one of the first implementation-phase software institutes to be awarded through NSF's Software Infrastructure for Sustained Innovation (SI2) program. SGCI has been designed as a service organization to address challenges by offering services to and building community among the research communities developing gateways. An application to be an XSEDE level two service provider is planned. The Institute's five-component design is the result of several years of studies, including many focus groups and a 5,000-person survey of the research community. This talk will describe SGCI's offerings and how they might benefit your work.
October 18, 2016
Towards Large-scale Genomics, Transcriptomics, and Metagenomics for All
Presenter(s): Philip Blood (PSC)
Principal Investigator(s): Noushin Ghaffari (Texas A&M) Ping Ma (U. Georgia) James Taylor (Johns Hopkins)
Although increasing numbers of researchers in genomics and related disciplines are utilizing advanced cyberinfrastructure for their work, these still represent a relatively small fraction of the biologists who could benefit from access to the latest genomics tools backed by large-scale computing resources. Rapid advances in these fields have caused an explosion of tools and algorithms that present a dizzying array of constantly changing options. Hence, even for scientists who are adept at using advanced computing infrastructure, it is challenging to determine the optimal mix of tools and employ these effectively to analyze large genomic data sets. In this talk I will highlight several XSEDE ECSS projects aimed at tackling aspects of these problems, both through formal ECSS collaborations and the "Novel and Innovative Projects" (NIP) arm of ECSS. These projects include the development of a pipeline for high-quality transcriptome analysis based on well-characterized RNA Sequencing Quality Control (SEQC) datasets, making memory-hungry sequence assembly tools available through the Galaxy XSEDE Gateway (usegalaxy.org), enabling large-scale analysis of human microbiome data, and facilitating the Critical Assessment of Metagenome Interpretation (CAMI: http://www.cami-challenge.org/).
Petascale DNS Using the Fast Poisson Solver PSH3D
Presenter(s): Darren Adams (NCSA)
Principal Investigator(s): Antonio Ferrante (U Wash)
Direct numerical simulation (DNS) of high Reynolds number (Re = O(105)) turbulent flows requires computational meshes of O(1012) grid points. Thus, DNS requires the use of petascale supercomputers. DNS often requires the solution of a Helmholtz (or Poisson) equation for pressure, which constitutes the bottleneck of the solver. We have developed and implemented a parallel solver of the Helmholtz equation in 3D called petascale Helmholtz 3D (PSH3D). The numerical method underlying PSH3D combines a parallel 2D Fast Fourier transform (P2DFFT) and a parallel linear solver (PLS). Our numerical results show that PSH3D scales up to at least 262,144 cores. PSH3D has a peak performance 6× faster than 3D FFT-based methods (e.g., P3DFFT) when used with the partial-global optimization. We have verified that the use of PSH3D with the partial-global optimization in our DNS solver does not reduce the accuracy of the numerical solution when tested for the Taylor-Green vortex flow.
September 20, 2016
Integrating Scientific Tools and Web Portals
Presenter(s): Kevin (Feng) Chen (TACC)
Principal Investigator(s): Carol X. Song (Purdue) Ritu Arora (TACC)
Abstract: Diagrid is powered by the HUBzero® software developed at Purdue University. It is specifically designed to help a scientific community share resources and work together with one another. The Diagrid Science as a Service platform allows for easy web-based access to software applications used by thousands of researchers around the world. In today's ECSS symposium, Dr. Kevin Chen will discuss the development on scientific tools leveraging Diagrid web portal and XSEDE HPC resources.
System-level Checkpoint-Restart with DMTCP
Presenter(s): Jerome Vienne (TACC)
Principal Investigator(s): Gene Cooperman (Northeastern University)
DMTCP (Distributed MultiThreaded CheckPointing) is a software package used to checkpoint-restart applications. The primary purpose of checkpointing in HPC is achieving fault tolerance. If a computation fails, whether for reasons of hardware failure or temporary software failure, then the user restarts the computation from a previous checkpoint. This presentation highlights work on ECSS project with the team that develops it. The initial purpose of the ECSS project was to provide support to extend the scalability of DMTCP but it ended to be more than that. During the presentation, I will introduce DMTCP and explain how it can be used to checkpoint-restart and debug a batch session, checkpoint OpenSHMEM implementations and large scale experiments running on InfiniBand clusters. All these points brought to different challenges that were solved during this ECSS project. This collaboration led to papers presented at XSEDE'16, OpenSHMEM 2016 and IEEE ICPADS 2016.
August 16, 2016
Re-presenting Large Image Collections for Data Mining and Analysis
Presenter(s): Paul Rodriguez (SDSC)
Principal Investigator(s): Elizabeth Wuerffel (Valparaiso University) Alison Langmead (University of Pittsburgh)
I will discuss two NIP/ECSS projects that both involve primarily image analysis in the context of digital humanities. (Image Analysis of Rural Photography, PI Wuerffel; Decomposing Bodies, [aka Image Analysis of Bertillon Prison Cards], PI Langmead). Both of them are superficially about taking old B&W photograph collections and 'digitizing' them. In a general sense, the goal is to re-represent the data so that the digital humanist can perform particular socio/historical/artistic/cultural analysees. In a more practical sense, the goal is to extract feature from the images and metadata and provide infrastructure support for analysis. Part of our challenge is to line up these two goals.
I will also discuss the technical and programmatic aspects, mostly for my own pieces of the projects. Although the processes, project logistics, and infrastructure are very similar between projects, the actual feature extraction code and data products have little overlap - which is due to the nature of the image data themselves. Feature extraction for both projects primarily involve an assembly of techniques/tools that are in open source packages, where the trickier aspects require coming up with good strategies for applying techniques, evaluating how well they work on this data, and exploring possible methods that might be useful to the user.
Experiences running Dynamic Traffic Assignment Simulations at scale using HPC Infrastructure
Presenter(s): Amit Gupta (TACC)
Principal Investigator(s): Natalia Ruiz Juri (UT)
Dynamic Traffic Assignment (DTA) simulations form an important analysis tool for Transportation researchers in attempting to model complex interactions between travelers and transportation infrastructure. These simulation frameworks are complex to develop, maintain and extend. VISTA is a widely used Transportation Simulation framework providing Dynamic Traffic Assignment. I discuss our experiences in scaling VISTA on the Stampede system as an exemplar of how HPC infrastructure and tools can augment analysis workflows in transportation research by significantly speeding up simulation experiments. I also discuss some challenges and tradeoffs in enabling DTA frameworks for use in HPC environments and also directions for continuing/future work under ECSS support.
June 21, 2016
SeedMe platform: Enabling scriptable data sharing
Presenter(s): Amit Chourasia (SDSC)
Abstract Most scientific computation and analyses create important transient data and preliminary results. Quick and effective access and assessments of this data is necessary for efficient use of researchers time and computation resources, but this process is complicated when a large collaborating team is geographically dispersed and/or some team members do not have direct access to the computation resource and output data. Current methods for sharing and assessing transient data and preliminary results are cumbersome, labor intensive, and largely unsupported by useful tools and procedures. Each research team is forced to create their own ad hoc procedures to push results from system to system, and user to user, to guide the next step in their research.
In this talk we introduce the SeedMe platform which provides a web-based cyberinfrastructure to enable easy sharing and streaming of transient data and preliminary results directly from computing resources to a variety of platforms, from mobile devices to workstations. The SeedMe platform is open to all researchers and provides web browser based as well scriptable tools for easy integration with ad hoc computation workflows. The talk will also briefly discuss applications and uses cases that may be relevant for ECSS and Science Gateway projects.
Biography Amit Chourasia is a Sr. Visualization Scientist at the San Diego Supercomputer Center (SDSC), UC San Diego. He leads the Visualization group where his work is focused on leading the research, development and application of software tools and techniques for visualization. Key area of his work is to develop methods to represent data in a visual form that is clear, succinct and accurate (a challenging yet very exciting endeavor). Data sharing is also at a forefront of his interests, to this end he is developed a web based infrastructure to enable this important and at times critical gap in scientific process via the SeedMe project.
May 17, 2016
Turning on Performance in LAMMPS Molecular Dynamics
Presenter(s): Kent Milfeld (TACC)
Principal Investigator(s): Peter Koenig (Procter & Gamble)
LAMMPS is a large-package atomistic and molecular dynamics simulator. Through the Industrial Challenge Program TACC supported Peter Koenig (PI) in using LAMMPS on the Stampede system. The object of the program was to create atomistic and particle simulations that could be used to determine micellar properties to confirm and replace experiments that develop rheology models for mixing, filling, and product performance predictions. The presentation will focus on the support work: LAMMPS optimizations, which included a few code changes, a description on adding new Classes (styles) for modifying interactions, and other efforts supporting efficient use of the Stampede system
Curation en masse: Exploration of the Quality of Video Collections
Presenter(s): Anne Bowen (TACC)
Principal Investigator(s): Alan Bovik (UT)
TACC provided support for Alan Bovik (Laboratory for Image and Video Engineering at UT Austin) to assess the use of automatic quality assessment algorithms at a large scale for museum digital video collections. This project involved developing a visual analysis tool and workflow for massive video quality assessment on TACC systems using the BRISQUE algorithm. The presentation will present an overview of the Quality Assessment workflow, and specifically focus on the challenges we encountered with using BRISQUE (and non-referential quality assessment algorithms in general) on museum collections. These challenges prompted the development of the visual analysis tool to assist with interpretation of the results.
April 19, 2016
How to Tune and Extract Higher Performance with MVAPICH2 Libraries
Presenter(s): Dhabaleswar K.(DK) Panda (Ohio State)
The Ohio State University MVAPICH2 libraries support the latest MPI 3.1 standard and deliver high performance, scalability and fault tolerance for high-end computing systems using InfiniBand, Omni-Path, 10-40 GigE/iWARP and RoCE (V1 and V2) networking technologies. MVAPICH2-GDR library uses novel designs to exploit the cutting-edge GPUDirect technology to provide high performance for MPI applications on systems with NVIDIA-GPUs. These libraries have multiple features, parameters and knobs to optimize the performance on modern systems. However, many users are not fully-aware of all these features, optimization and tuning techniques. This talk is aimed to address these concerns and provide a set of concrete guidelines to XSEDE users to boost performance of their applications. We will start with an overview of the MVAPICH2 libraries and their features and optimized designs. Next, we will provide an in-depth overview of the runtime optimizations and tuning flexibility. We will demonstrate how you can tune and optimize these libraries to fit the needs of your application on a given system. Using a set of `Best Practice' examples, we will highlight the impact of tuning and optimizations on a set of common XSEDE applications including Amber, Lulesh, Hoomdblue, and MILC.
Bio --------- DK Panda is a Professor and University Distinguished Scholar of Computer Science and Engineering at the Ohio State University. He has published over 350 papers in the area of high-end computing and networking. The MVAPICH2 (High Performance MPI and PGAS over InfiniBand, iWARP and RoCE) libraries, designed and developed by his research group (http://mvapich.cse.ohio-state.edu), are currently being used by more than 2,550 organizations worldwide (in 79 countries). More than 360,000 downloads of this software have taken place from the project's site. This software is empowering several InfiniBand clusters (including the 10th, 13th and 25th ranked ones) in the TOP500 list. The RDMA packages for Apache Spark, Apache Hadoop and Memcached together with OSU HiBD benchmarks from his group (http://hibd.cse.ohio-state.edu) are also publicly available. These libraries are currently being used by more than 160 organizations in 22 countries. More than 15,900 downloads of these libraries have taken place. He is an IEEE Fellow. More details about Prof. Panda are available at http://www.cse.ohio-state.edu/~panda
March 15, 2016
XDMoD(XD Metrics Service)
Presenter(s): Thomas Furlani (University at Buffalo, SUNY)
The University at Buffalo's XDMoD tool provides for the comprehensive management of HPC systems, including the ability to provide performance data for all jobs running on a cluster. Using the XDMoD Job Viewer, system support personnel can readily identify poorly performing jobs - with the end goal of working the the end user to improve performance.
In this presentation, we will begin with a brief PowerPoint presentation on XDMoD with an emphasis on the Job Viewer tab. This will be followed by a live demo that utilizes the Job Viewer within XDMoD to analyze various XSEDE jobs. The demo will be interactive - allowing ECSS staff to help guide the demo.
Link to the recorded presentation.
May require download of proprietary software to view video
The XDMoD team is interested in collecting feedback on usability. David LaVergne is conducting one-on-one interviews with users regarding their use of the current interface (mainly the Usage and Metrics Explorer tabs, as well as the new Job Viewer tab). As user support folks, I'd like to know generally what information you find the most (and least) useful, and how well the current interface (and a proposed redesign) supports that. Please contact him if interested, a small gift is involved!