학술논문
ASCR/HEP Exascale Requirements Review Report
Document Type
article
Author
Habib, Salman; Roser, Robert; Gerber, Richard; Antypas, Katie; Riley, Katherine; Williams, Tim; Wells, Jack; Straatsma, Tjerk; Almgren, A; Amundson, J; Bailey, S; Bard, D; Bloom, K; Bockelman, B; Borgland, A; Borrill, J; Boughezal, R; Brower, R; Cowan, B; Finkel, H; Frontiere, N; Fuess, S; Ge, L; Gnedin, N; Gottlieb, S; Gutsche, O; Han, T; Heitmann, K; Hoeche, S; Ko, K; Kononenko, O; LeCompte, T; Li, Z; Lukic, Z; Mori, W; Nugent, P; Ng, C-K; Oleynik, G; O'Shea, B; Padmanabhan, N; Petravick, D; Petriello, FJ; Power, J; Qiang, J; Reina, L; Rizzo, TJ; Ryne, R; Schram, M; Spentzouris, P; Toussaint, D; Vay, J-L; Viren, B; Wurthwein, F; Xiao, L
Source
Subject
Language
Abstract
This draft report summarizes and details the findings, results, andrecommendations derived from the ASCR/HEP Exascale Requirements Review meetingheld in June, 2015. The main conclusions are as follows. 1) Larger, morecapable computing and data facilities are needed to support HEP science goalsin all three frontiers: Energy, Intensity, and Cosmic. The expected scale ofthe demand at the 2025 timescale is at least two orders of magnitude -- and insome cases greater -- than that available currently. 2) The growth rate of dataproduced by simulations is overwhelming the current ability, of both facilitiesand researchers, to store and analyze it. Additional resources and newtechniques for data analysis are urgently needed. 3) Data rates and volumesfrom HEP experimental facilities are also straining the ability to store andanalyze large and complex data volumes. Appropriately configuredleadership-class facilities can play a transformational role in enablingscientific discovery from these datasets. 4) A close integration of HPCsimulation and data analysis will aid greatly in interpreting results from HEPexperiments. Such an integration will minimize data movement and facilitateinterdependent workflows. 5) Long-range planning between HEP and ASCR will berequired to meet HEP's research needs. To best use ASCR HPC resources theexperimental HEP program needs a) an established long-term plan for access toASCR computational and data resources, b) an ability to map workflows onto HPCresources, c) the ability for ASCR facilities to accommodate workflows run bycollaborations that can have thousands of individual members, d) to transitioncodes to the next-generation HPC platforms that will be available at ASCRfacilities, e) to build up and train a workforce capable of developing andusing simulations and analysis to support HEP scientific research onnext-generation systems.