Sign up for our newsletter and get the latest HPC news and analysis.
Send me information from insideHPC:


Containers Using Singularity on HPC

Abhinav Thota, from Indiana University gave this talk at the 2018 Swiss HPC Conference. “Container use is becoming more widespread in the HPC field. There are various reasons for this, including the broadening of the user base and applications of HPC. One of the popular container tools on HPC is Singularity, an open source project coming out of the Berkeley Lab. In this talk, we will introduce Singularity, discuss how users of Indiana University are using it and share our experience supporting it. This talk will include a brief demonstration as well.”

Call for Submissions: SC18 Workshop on Reproducibility

Over at the SC18 Blog, Stephen Lien Harrell from Purdue writes that the conference will host will host a workshop on the hot topic of Reproducibility. Their Call for Submissions is out with a deadline of August 19, 2018. ”
The Systems Professionals Workshop is a platform for discussing the unique challenges and developing the state of the practice for the HPC systems community. The program committee is soliciting submissions that address the best practices of building and operating high performance systems with an emphasis on reproducible solutions that can be implemented by systems staff at other institutions.”

Using the Titan Supercomputer to Develop 50,000 Years of Flood Risk Scenarios

Dag Lohmann from KatRisk gave this talk at the HPC User Forum in Tucson. “In 2012, a small Berkeley, California, startup called KatRisk set out to improve the quality of worldwide flood risk maps. The team wanted to create large-scale, high-resolution maps to help insurance companies evaluate flood risk on the scale of city blocks and buildings, something that had never been done. Through the OLCF’s industrial partnership program, KatRisk received 5 million processor hours on Titan.”

Fujitsu Upgrades RAIDEN at RIKEN Center for Advanced Intelligence Project

Fujitsu reports that the company has significantly boosted the performance of the RAIDEN supercompuer. RAIDEN is a computer system for artificial intelligence research originally deployed in 2017 at the RIKEN Center for Advanced Intelligence Project (AIP Center). “The upgraded RAIDEN has increased its performance by a considerable margin, moving from an initial total theoretical computational performance of 4 AI Petaflops to 54 AI Petaflops, placing it in the top tier of Japan’s systems. In having built this system, Fujitsu demonstrates its commitment to support cutting-edge AI research in Japan.”

Why UIUC Built HPC Application Containers for NVIDIA GPU Cloud

In this video from the GPU Technology Conference, John Stone from the University of Illinois describes how container technology in the NVIDIA GPU Cloud help the University distribute accelerated applications for science and engineering. “Containers are a way of packaging up an application and all of its dependencies in such a way that you can install them collectively on a cloud instance or a workstation or a compute node. And it doesn’t require the typical amount of system administration skills and involvement to put one of these containers on a machine.”

Video: HPC Use for Earthquake Research

Christine Goulet from the Southern California Earthquake Center gave this talk at the HPC User Forum in Tucson. “SCEC coordinates fundamental research on earthquake processes using Southern California as its principal natural laboratory. The SCEC community advances earthquake system science through synthesizing knowledge of earthquake phenomena through physics-based modeling, including system-level hazard modeling and communicating our understanding of seismic hazards to reduce earthquake risk and promote community resilience.”

Intel Open Sources nGraph Deep Neural Network model for Multiple Devices

Over at Intel, Scott Cyphers writes that the company has open-sourced nGraph, a framework-neutral Deep Neural Network (DNN) model compiler that can target a variety of devices. With nGraph, data scientists can focus on data science rather than worrying about how to adapt their DNN models to train and run efficiently on different devices. Continue reading below for highlights of our engineering challenges and design decisions, and see GitHub, our documentation, and our SysML paper for additional details.

JUWELS Supercomputer in Germany to be based on Modular Supercomputing

“The supercomputer JUQUEEN, the one-time reigning power in Europe’s high-performance computing industry, is ceding its place to its successor, the Jülich Wizard for European Leadership Science. Called JUWELS for short, the supercomputer is the culmination of the joint efforts of more than 16 European partners in the EU-funded DEEP projects since 2011. Once completed, JUWELS will consist of three fully integrated modules able to carry out demanding simulations and scientific tasks.”

Universities step up to Cloud Bursting

In this special guest feature, Mahesh Pancholi from OCF writes that many of universities are now engaging in cloud bursting and are regularly taking advantage of public cloud infrastructures that are widely available from large companies like Amazon, Google and Microsoft. “By bursting into the public cloud, the university can offer the latest and greatest technologies as part of its Research Computing Service for all its researchers.”

Charliecloud: Unprivileged Containers for User-Defined Software Stacks

“What if I told you there was a way to allow your customers and colleagues to run their HPC jobs inside the Docker containers they’re already creating? or an easily learned, easily employed method for consistently reproducing a particular application environment across numerous Linux distributions and platforms? There is. In this talk/tutorial session, we’ll explore the problem domain and all the previous solutions, and then we’ll discuss and demo Charliecloud, a simple, streamlined container runtime that fills the gap between Docker and HPC — without requiring HPC Admins to lift a finger!”