The Exascale Computing Project

Lori Diachin (LLNL) Deputy Director, Exascale Computing Project

SC18 Dallas, TX November 14, 2018

exascaleproject.org DOE Exascale Program: The Exascale Computing Initiative (ECI)

ECI US DOE (SC) and National partners Nuclear Security Administration (NNSA) Exascale Selected program office Computing Project Accelerate R&D, acquisition, and deployment to application ECI development (ECP) deliver exascale computing capability to DOE (BER, BES) mission national labs by the early- to mid-2020s

Delivery of an enduring and capable exascale Exascale system ECI computing capability for use by a wide range procurement projects focus of applications of importance to DOE and the US ALCF-3 () OLCF-5 (Frontier) ASC ATS-4 (El Capitan)

2 Three Major Components of the ECI ECP enables US revolutions in technology development; scientific discovery; healthcare; energy, economic, and national security

ECP ECP mission vision

Develop exascale-ready applications and Deliver exascale simulation and solutions that address currently intractable data science innovations and problems of strategic importance and solutions to national problems national interest. that enhance US economic competitiveness, change our quality Create and deploy an expanded and of life, and strengthen our national vertically integrated software stack on security. DOE HPC exascale and pre-exascale systems, defining the enduring US exascale ecosystem.

Deliver US HPC vendor technology advances and deploy ECP products to DOE HPC pre-exascale and exascale systems.

3 Department of Energy (DOE) Roadmap to Exascale Systems An impressive and productive lineup of accelerated node systems supporting DOE’s mission

Pre-Exascale [Aggregate Linpack (Rmax) = 323 PF!] First U.S. Exascale Computers 2012 2016 2018 2020 2021-2023

Titan (9) (1) ORNL ORNL /AMD/NVIDIA ORNL TBD IBM/NVIDIA

Aurora Mira (21) Theta (24) ANL ANL ANL /Cray IBM BG/Q Cray/Intel KNL NERSC-9 Cori (12) Perlmutter LBNL LBNL Cray/Intel Xeon/KNL Cray/AMD/NVIDIA

Sequoia (10) LLNL Trinity (6) Sierra (2) TBD LLNL LLNL IBM BG/Q LANL/SNL LANL/SNL Cray/Intel Xeon/KNL IBM/NVIDIA 4 TBD The three technical areas in ECP have the necessary components to meet national goals Performant mission and science applications @ scale Foster application Ease Diverse HPC development of use architectures leadership

Application Development (AD) Software Hardware Technology (ST) and Integration (HI) Develop and enhance the predictive Produce expanded and vertically Integrated delivery of ECP capability of applications critical to integrated software stack to achieve products on targeted systems at the DOE full potential of exascale computing leading DOE computing facilities

6 vendors supported 89 unique software by PathForward products spanning focused on memory, 25 applications ranging from national programming models node, connectivity security, to energy, earth systems, and runtimes, math and reliability economic security, materials, and data libraries, data and advancements; visualization deployment to 5 facilities Back-up

6 ECP applications target national problems in six areas

National security Energy security Economic security Scientific discovery Earth system Health care Next-generation, Turbine wind plant Additive Cosmological probe Accurate regional Accelerate stockpile efficiency manufacturing of the standard impact assessments and translate stewardship codes Design and of qualifiable model of particle in Earth system cancer research Reentry-vehicle- commercialization metal parts physics models environment of SMRs Urban planning Validate Stress-resistant simulation Nuclear fission Reliable and fundamental laws of crop analysis and Multi-physics and fusion reactor efficient planning nature catalytic conversion science simulations materials design of the power grid Plasma wakefield of biomass-derived alcohols of high-energy Subsurface use Seismic hazard accelerator design density physics for carbon capture, risk assessment Light source- Metagenomics conditions petroleum enabled analysis of for analysis of extraction, waste protein and biogeochemical disposal molecular structure cycles, climate and design change, High-efficiency, environmental low-emission Find, predict, remediation combustion engine and control and gas turbine materials and design properties Scale up of clean Predict and control fossil fuel stable ITER combustion operational Biofuel catalyst performance design Demystify origin of chemical elements

7 Software Development Kits Progress: Leadership in place, Spack packaging making rapid progress

ECP software projects Each project to define (potentially ≥2) release vectors More projects Fewer projects OpenHPC SDKs Math SDK Direct2Facility Reusable software libraries Potential exit strategy Platform-specific software embedded in applications; Tools SDK for binary distributions in support of a specified cohesive/interdependent PM&RT SDK • Target similar software to 2021–2023 exascale system libraries released as sets existing OpenHPC stack DataViz SDK • Software exclusively modeled on xSDK • Develop super-scalable supporting a specific platform release targeting higher end • System software, some tools systems • Regular coordinated and runtimes releases • Hierarchical collection built on Spack • Products may belong to >1 SDK Leadership Team: Decades of Software Experience SDK based on dependences - Jim Willenbring – SDK Coordinator and Release Manager • Establish community policies - Sameer Shende – Programming Models & Runtimes for library development - Bart Miller – Development Tools • Apply Continuous Integration and other robust testing - Lois McInnes – Math Libraries practices - Chuck Atkins - Data & Viz 8 PathForward funds 6 US HPC companies to accelerate technologies to maximize the energy efficiency and overall performance of future

• Accelerate critical early hardware R&D leading to 3–5 viable exascale system • Advanced Micro Devices (AMD) designs for DOE Facilities • Cray Inc. (CRAY) • 3-year program ending early in 2020 • Hewlett Packard Enterprise (HPE) • Examples of work funded include: • International Business Machines (IBM) a) innovative memory architectures b) higher-speed interconnects • Intel Corp. (Intel) c) improved system reliability • NVIDIA Corp. (NVIDIA) d) approaches for increasing computing power without prohibitive increases in energy demand

9 LLNL Auspices and Disclaimer

Prepared by LLNL under Contract DE-AC52-07NA27344. This document was prepared as an account of work sponsored by an agency of the government. Neither the United States government nor Lawrence Livermore National Security, LLC, nor any of their employees makes any warranty, expressed or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or Lawrence Livermore National Security, LLC. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or Lawrence Livermore National Security, LLC, and shall not be used for advertising or product endorsement purposes.

10