The third of three new exa-scale computers set to be purchased and built by Secretary Rick Perry and the Department of Energy. Robin Goldstone, a solution architect at LLNL, gave an interview at the Red Hat Summit 2019. She suggested el capitan will be 10x the performance of Sierra (which is 125 petaflops, and "so El Capitan's targeted to be 1.2, maybe 1.5 exaFLOPS, or even more. Again, that's peak performance, it doesn't necessarily translate into what our applications can get out of the platform." In concept, we can use this to "push a workload through 10 times faster or we can look at a simulation that's 10 times more resolved" (better detail). There will be an El Capitan Center of Excellence at LLNL (deployment and production use from 2023 - 2038). It should focus on AI and machine learning, and advancing the goal of intelligent simulation or cognitive computing. El Capitan (ATS-4) is the successor to Sierra, and will be succeeded by (ATS-6) in 2027-2028 timeframe. In March 2019's DoE FY2020 Congressional Budget Justification: Highlights of the FY 2020 Budget:
LLNL needs more computing capability to achieve and advance its mission. "More computing capability is needed." (Original in bold.) Good one pagers on the DoE Exascale Computing Project (ECP) (thank you Rick Stevens, Argonne National Laboratory & The University of Chicago): Sources:
https://video.cube365.net/v/Plqw-LotHfo https://asc.llnl.gov/content/assets/docs/ElCap_IntelligentSimCOE_whitepaper.pdf https://en.wikichip.org/wiki/supercomputers/ats-4 https://www.energy.gov/sites/prod/files/2019/04/f62/doe-fy2020-budget-volume-1.pdf http://www.sci.utah.edu/~chris/Rick-Stevens-DSCI-Lecture-2019.pdf https://exascale.llnl.gov/ https://www.exascaleproject.org/
0 Comments
By Jeffrey Cohen, @chicago_quantum, Founder of Chicago Quantum. June 10, 2019 Seven facts about Frontier (from the sources): - Weighs over a million pounds - Performance equals the world's top 160 fastest supercomputers. - Performance of 1.5 exaflops, equals 1.5 quintillion operations per second, equals 1.5 x billion x billion operations per second. - Partnership between Oak Ridge National Laboratory (ORNL), U.S. Department of Energy, Cray and AMD, with a Cray contract including $6oom for a CoE, early delivery systems, the main Frontier system, and multi-year system support. - Cray Programming Environment (Cray PE) will be enhanced in 3 ways: enhanced high-level software development environment, integrated with AMD technology with Cray Slingshot to take better advantage of the hardware, and integrated with a full machine-learning software stack. - Establish a Center of Excellence by Cray and Oak Ridge National Lab Frontier will be based on Cray's architecture, AMD's CPU and GPU technologies. - Lawrence Livermore Lab will have an exascale computer too (pre-announced), and Argonne National Lab's Aurora @ 1.0 exaflops was recently announced (Intel and Cray for $500M). Sources: https://www.cray.com/resources/how-frontier-measures-up https://www.cray.com/blog/doe-and-cray-announce-exascale-supercomputer-frontier/#comment-757096 http://investors.cray.com/phoenix.zhtml?c=98390&p=irol-newsArticle&ID=2397457 Implications: While Energy Secretary Rick Perry discusses how we will 'win' the race to develop quantum computers, he is ensuring the U.S. has access to exascale classical computing too. |
Jeff CohenStrategic IT Management Consultant with a strong interest in Quantum Computing. Consulting for 29 years and this looks as interesting as cloud computing was in 2010. Archives
February 2021
|