Extreme Scale Deep Learning at NERSC
Thorsten Kurth from LBNL gave this talk at the PASC18 conference. "We present various studies on very large scale distributed deep learning on HPC systems including the ~10k node Intel Xeon-Phi-based Cori system at NERSC. We explore CNN classification architectures and generative adversarial networks for HEP problems using large images corresponding to full LHC detectors and high-resolution cosmology convergence maps." [READ MORE...]
Featured Stories
Legendary Magnum InfiniBand Switch comes to Computer History Museum
The Computer History Museum in Silicon Valley recently added a piece of TACC's history into their permanent historical collection — sealing its place as a milestone in computing.
We're always searching around the world for new, interesting, and important computing objects," said Dag Spicer, senior curator at the Computer History Museum. "TACC's Sun Microsystems 2007 Magnum switch was a critical part of high-performance computing at that time in history. The TACC switch was the largest of its class and is an example of InfiniBand technology, which has long been a key enabling technology for high performance computing." [READ MORE...]
ISC 2018: NVIDIA DGX-2 -- The World’s Most Powerful AI System on Display
In this video, Satinder Nijjar from NVIDIA describes the new DGX-2 GPU supercomputer. "Experience new levels of AI speed and scale with NVIDIA DGX-2, the first 2 petaFLOPS system that combines 16 fully interconnected GPUs for 10X the deep learning performance. It’s powered by NVIDIA DGX software and a scalable architecture built on NVIDIA NVSwitch, so you can take on the world’s most complex AI challenges." [READ MORE...]
DDN Steps Up to HPC & AI Workloads at ISC 2018
In this video from ISC 2018, James Coomer from DDN describes the company's latest high performance storage technologies for AI and HPC workloads. "Attendees at ISC 2018 learned how organizations around the world are leveraging DDN’s people, technology, performance and innovation to achieve their greatest visions and make revolutionary insights and discoveries! Designed, optimized and right-sized for Commercial HPC, Higher Education and Exascale Computing, our full range of DDN products and solutions are changing the landscape of HPC and delivering the most value with the greatest operational efficiency." [READ MORE...]
Featured Resource
PLASTER: A Framework for Deep Learning Performance
Both hardware and software advances in deep learning (DL), a type of ML, appear to be catalysts for the early stages of a phenomenal AI growth trend. Download the new white paper from NVIDIA that addresses the challenges described in PLASTER, which is important in any deep learning solution, and it is especially useful for developing and delivering the inference engines underpinning AI-based services.
Industry Perspectives
Podcast: Evolving MPI for Exascale Applications
In this episode of Let's Talk Exascale, Pavan Balaji and Ken Raffenetti describe their efforts to help MPI, the de facto programming model for parallel computing, run as efficiently as possible on exascale systems. "We need to look at a lot of key technical challenges, like performance and scalability, when we go up to this scale of machines. Performance is one of the biggest things that people look at. Aspects with respect to heterogeneity become important." [Read More...]
Radio Free HPC Trip Report from ISC 2018
In this podcast, the Radio Free HPC team offers up a Trip Report from ISC 2018 in Frankfurt. It was a whirlwind week for news with a new USA machine on the TOP500, but the other big news centered around the convergence of HPC & AI. This common theme was all over the show floor, with use cases on display in dozens of exhibits. [Read More...]
Featured from insideBIGDATA
- Data Science at Microsoft – Interviews with PractitionersIn this technical brief I wanted to pass along some great resources in support how data scientists approach their profession and illustrate the kind of background a typical data scientist might have to become successful. insideBIGDATA previously featured four compelling podcast interviews with Microsoft data scientists.
News from insideBIGDATA
- Big Data Day LA 2018
- Data Science at Microsoft – Interviews with Practitioners
- How Retail Can Thrive Using AI as ‘The Human Touch’
- Interview: Bill Waid, VP and General Manager, FICO Decision Management Suite
- Interview: Jamie Engesser, VP Product Mangement at Hortonworks
- Best of arXiv.org for AI, Machine Learning, and Deep Learning – June 2018
- Facing Your Data Framework
Editor’s Choice
HPE Teams with WekaIO for HPC and Machine Learning
In this video from the HPE booth at ISC 2018, Barbara Murphy from WekaIO describes how the company's fast parallel file system for NVMe devices speeds HPC and Machine Learning workloads. "You invested in the best infrastructure to run your AI and data intensive applications, so don't let data accessibility be the bottleneck to your productivity gains. To maximize your investments and resolve the GPU and CPU starvation problem, you need a shared file system that ensures data is available to the applications. WekaIO delivers all the bandwidth you need, so your applications never have to wait for data." [Read More...]
Unravelling Earthquake Dynamics through Extreme-Scale Multiphysics Simulations
Alice-Agnes Gabriel gave this talk at the PASC18 conference. "Earthquakes are highly non-linear multiscale problems, encapsulating geometry and rheology of faults within the Earth’s crust torn apart by propagating shear fracture and emanating seismic wave radiation. This talk will focus on using physics-based scenarios, modern numerical methods and hardware specific optimizations to shed light on the dynamics, and severity, of earthquake behavior." [Read More...]
Announcing the new Intel Select Solution for Professional Visualization
In this video from ISC 2018, Jim Jeffers from Intel describes the new Intel Select Solution for Professional Visualization, an easily-deployed, Intel-optimized system reference architecture that is purpose-built to meet the demands of today’s most complex data explosion challenges. ""Visualization is a key component in advanced computing that allows systems to deliver greater insights with faster turnaround when using large-scale data sets." [Read More...]
Virtualizing AI & HPC Workloads with vSphere
In this video from the Dell EMC HPC Community Meeting, Josh Simons from VMware describes why more customers are moving their HPC & AI workloads to virtualized environments using vSphere. "With VMware, you can capture the benefits of virtualization for HPC workloads while delivering performance that is comparable to bare-metal. Our approach to virtualizing HPC adds a level of flexibility, operational efficiency, agility and security that cannot be achieved in bare-metal environments—enabling faster time to insights and discovery." [Read More...]
Let's Talk Exascale: Optimizing I/O at the ADIOS Project
In this episode of Let's Talk Exascale, researchers from the ADIOS project describe how they are optimizing I/O on exascale architectures and making the code easily maintainable, sustainable, and extensible, while ensuring its performance and scalability. "The Adaptable I/O System (ADIOS) project in the ECP supports exascale applications by addressing their data management and in situ analysis needs." [Read More...]
