A-Z Index | Directory | Careers

Berkeley Lab’s Leadership in Deep Learning, Exascale Computing and Beyond on Tap at SC17

November 6, 2017

SC17logo 300x300

Lawrence Berkeley National Laboratory staff will be sharing their expertise reflecting the lab’s work in pushing the envelope to exascale computing and beyond, as well as Berkeley Lab’s leadership in deep learning.

In the area of deep learning, lab staff co-authored papers on “Deep Learning at 15PF: Supervised and Semi-Supervised Classification for Scientific Data” and “Scaling Deep Learning on GPU and Knights Landing Clusters.” Prabhat will also give a plenary talk on Sunday, Nov. 12 at the Intel Developer Conference held in conjunction with SC17.

In the exascale arena, Kathy Yelick will be part of a presentation on recent accomplishments of DOE’s Exascale Computing Project at 2:30 p.m. Tuesday, Nov. 14 in the DOE booth (613). Two simulations running in the DOE booth were created using adaptive mesh refinement software developed by AMReX, the ECP Block-Structured Adaptive Mesh Refinement Co-Design Center. Lab staff also co-organized a Nov. 14 Birds-of-a-Feather session on Reconfigurable Computing in Exascale.”

Looking beyond exascale, in a DOE booth presentation at 3:15 p.m. Tuesday, Nov. 14, Jonathan Carter will discuss Berkeley Lab's Advanced Quantum-Enabled Simulation (AQuES) Testbed, highlighting recent and future developments in our sc-qubit platform, control electronics and coupling to classical computing. Examples of the project’s eight-qubit rings will be displayed in the booth. George Michelogiannakis will join a Nov. 16 panel discussion on “Post Moore Supercomputing,”

Lab staff will also share their expertise with the global HPC committee through other tutorials, technical papers and focused workshops in the SC17 conference technical program, as well as numerous demos in the DOE booth (613). Here are several examples of leading-edge presentations by Berkeley Lab experts:

The Roofline Model for Performance Tuning: Roofline Model developer Sam Williams will co-lead a Nov. 12 tutorial on “Performance Tuning of Scientific Codes with the Roofline Model” and host a demo on “Using the Roofline Model and Intel Advisor” at 2 p.m. Tuesday, Nov. 14 in the DOE booth (no. 613).

The Cori Burst Buffer: NERSC pioneered the deployment of an SSD “burst buffer” with the installation of Cori, using this added storage layer to accelerate I/O. NERSC’s Deborah Bard will co-present the “Getting Started with the Burst Buffer: Using DataWarp Technology” tutorial on Monday, Nov. 13, then host an hour-long “Hands-on with Cori's Burst Buffer,” demo at 4 p.m. Wednesday, Nov. 15, in the DOE booth (no. 613)  This will be followed by a roundtable discussion, also in the DOE booth.

Additionally, staff members will participate in a number of conference outreach programs for students and early career professionals.

Here’s a day-by-day guide to sessions featuring Berkeley Lab staff -- click on the links for more information:

Sunday, November 12

Tutorials

OpenMP Common Core: A “Hands-On” Exploration,” presenters include Alice Koniges, and Yun (Helen) He, NERSC;8:30 a.m. - 5 p.m.

Performance Tuning of Scientific Codes with the Roofline Model,” presenters include Tuomas S. Koskela, NERSC, and Samuel W. Williams, Computational Research Division; 8:30 a.m. - 12 p.m.

Workshops

7th Workshop on Python for High-Performance and Scientific Computing,” organizers include Rollin Thomas, NERSC; 9 a.m. - 5:30 p.m.

Eighth Annual Workshop for the Energy Efficient HPC Working Group,” organizers include Dale Sartor, Building Technology and Urban Systems Division; 9 a.m. - 5:30 p.m.

ISAV 2017: In Situ Infrastructures for Enabling Extreme-Scale Analysis and Visualization,” presenters include E. Wes Bethel and Gunther Weber, CRD; 9 a.m. - 5:30 p.m.

Women in HPC: Diversifying the HPC Community,” organizers include Rebecca Hartman-Baker, NERSC; 9 a.m. - 5:30 p.m.

Workshop for Open Source Supercomputing,” presenters include Farzad Fatollahi-Fard, David Donofrio and Anastasiia Butko, CRD.; 9 a.m. - 5:30 p.m.

Fourth SC Workshop on Best Practices for HPC Training,” presenters include Richard Gerber and Rebecca Hartman-Baker, NERSC; 2 - 5:30 p.m.

Monday, November 13

Tutorials

Getting Started with the Burst Buffer: Using DataWarp Technology,” presenters include Deborah Bard, NERSC; 8:30 a.m. - 12 p.m.

In Situ Analysis and Visualization with SENSEI,” presenters include E. Wes Bethel and Burlen Loring, CRD; 8:30 a.m. - 12 p.m.

Parallel I/O in Practice,” presenters include Katie Antypas, NERSC;8:30 a.m. - 5 p.m.

Container Computing for HPC and Scientific Workflows,” Lisa Gerhardt and Shane Canon, NERSC;1:30 - 5 p.m.

Interactive HPC: Using C++ and HPX Inside Jupyterhub to Write Performant Portable Parallel Code,” presenters include Alice Koniges and Bryce A. Lelbach, CRD; 1:30 - 5 p.m.

Workshop

PAW 2017: The 2nd Annual PGAS Applications Workshop,” presenters include Costin Iancu, CRD; 9 a.m. - 5:30 p.m.

Tuesday, November 14

Papers

Deep Learning at 15PF: Supervised and Semi-Supervised Classification for Scientific Data,” authors include Thorsten Kurth, Evan Racah, Tareq Malas, Wahid Bhimji, Jack Deslippe and Prabhat, NERSC; 10:30 - 11 a.m.

Optimizing the Query Performance of Block Index Through Data Analysis and I/O Modeling,” authors include Bin Dong and Kesheng “John”  Wu, CRD; 11:30 a.m. - 12 p.m.

Scaling Deep Learning on GPU and Knights Landing Clusters,” authors include Aydin Buluç, CRD; 11:30 a.m. - 12 p.m.

Galactos: Computing the 3-pt Anisotropic Correlation for 2 Billion Galaxies,” authors include  Brian Friesen, Brian Austin, Zachary Slepian, Deborah Bard, Jack Deslippe and Prabhat, NERSC; 4 - 4:30 p.m

Invited Talk

"TOP500 - Past, Present, Future,"presenters include Erich Strohmaier, CRD, and Horst Simon, Director's Office; 4:15 - 5 p.m.

Research Poster

"A novel feature-preserving spatial mapping for deep learning classification of RAS structures," authors are Thomas Corcoran, Rafael Zamora-Resendiz, Xinlian Liu and Silvia Crivelli, CRD; 8:30 a.m. - 5 p.m., special poster reception from 5:15 - 7p.m. Tuesday, Nov. 14.

Featured talk in DOE Booth (613)

"Advanced Quantum-Enabled Simulation (AQuES) Testbed," Jonathan Carter, CRD;  3:15 p.m.

Demos in DOE Booth (613)

“Using the Roofline Model and Intel Advisor,” Sam Williams, CRD;2 - 3 p.m.

“Calibers: A Bandwidth Calendaring Paradigm for Science Workflows,” Eric Poulyou, ESnet, 3 - 4 p.m.

Birds-of-a-Feather Sessions

Containers in HPC,” organizers include Shane Canon, NERSC, and Greg Kurtzer, IT; 5:15 - 7 p.m.

HPC Graph Toolkits and the GraphBLAS Forum,” organizers include Aydin Buluc, CRD; 5:15 - 7 p.m.

Reconfigurable Computing in Exascale,” organizers include David Donofrio, CRD; 5:15 - 7 p.m.

TOP500 Supercomputers,” organizers include Erich Strohmaier, CRD,and Horst Simon, Director’s Office; 5:15 - 7 p.m.

Usability, Scalability and Productivity on Many-Core Processors: Intel Xeon Phi,” organizers include Doug Doerfler, NERSC; 5:15 - 7 p.m.

Wednesday, November 15

Invited Talk

"Inference and Control in Routing Games," Alex Bayen, UC Berkeley and the Energy Analysis and Environmental Impacts Division; 4:15 - 5 p.m.

Research Poster

"A novel feature-preserving spatial mapping for deep learning classification of RAS structures," authors are Thomas Corcoran, Rafael Zamora-Resendiz, Xinlian Liu and Silvia Crivelli, CRD; 8:30 a.m. - 5 p.m.

Demos in DOE Booth (613)

“Calibers: A Bandwidth Calendaring Paradigm for Science Workflows,” Eric Poulyou, ESnet, 10 - 11 a.m.

"Scaling Celeste to 1.5PF wityh Julia," Keno fischer (Julia Computing) and Prabhat, NERSC, 12 - 1 p.m.

Shifter: Container Computing for HPC,” Doug Jacobson and Shane Canon, NERSC; 2 - 3 p.m.

“Hands-on with Cori's Burst Buffer,” Deborah Bard, NERSC; 4 - 5 p.m.

Birds-of-a-Feather Sessions

Women in HPC: Non-Traditional Paths to HPC and How They Can and Do Enrich the Field,” organizers include Rebecca Hartman-Baker, NERSC; 12:15 - 1:15 p.m.

The Green500: Trends in Energy-Efficient Supercomputing,” organizers include Erich Strohmaier, CRD; 5:15 - 7 p.m.

Thursday, November 16

Research Poster

"A novel feature-preserving spatial mapping for deep learning classification of RAS structures," authors are Thomas Corcoran, Rafael Zamora-Resendiz, Xinlian Liu and Silvia Crivelli, CRD; 8:30 a.m. - 5 p.m.

Demos in DOE Booth (613)

“Calibers: A Bandwidth Calendaring Paradigm for Science Workflows,” Eric Poulyou, ESnet; 12 - 1 p.m.

Panel Discussion

Post Moore Supercomputing,” panelists include George Michelogiannakis, CRD; 3:30 - 5 p.m.

Friday, November 17

Panel Discussions

HPC Software: Is “Cool Stuff” Really Incompatible with Sustainability?,” panelists include Rebecca Hartman-Baker, NERSC; 8:30 - 10 a.m.

Energy Efficiency Gains: Retrospectives and Perspectives,” panelists include John Shalf, CRD; 10:30 a.m. - 12 p.m.


About Computing Sciences at Berkeley Lab

High performance computing plays a critical role in scientific discovery, and researchers increasingly rely on advances in computer science, mathematics, computational science, data science, and large-scale computing and networking to increase our understanding of ourselves, our planet, and our universe. Berkeley Lab’s Computing Sciences Area researches, develops, and deploys new foundations, tools, and technologies to meet these needs and to advance research across a broad range of scientific disciplines.

Founded in 1931 on the belief that the biggest scientific challenges are best addressed by teams, Lawrence Berkeley National Laboratory and its scientists have been recognized with 13 Nobel Prizes. Today, Berkeley Lab researchers develop sustainable energy and environmental solutions, create useful new materials, advance the frontiers of computing, and probe the mysteries of life, matter, and the universe. Scientists from around the world rely on the Lab’s facilities for their own discovery science. Berkeley Lab is a multiprogram national laboratory, managed by the University of California for the U.S. Department of Energy’s Office of Science.

DOE’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States, and is working to address some of the most pressing challenges of our time. For more information, please visit energy.gov/science.