ABOUT
Aurora is the Argonne Leadership Computing Facility’s next big advancement in supercomputing. Designed in collaboration with industry leaders Intel and Cray and operated in support of open science as part of the U.S. Department of Energy’s Leadership Computing Facility, Aurora is a next-generation system for computational science and engineering.
Aurora exists for the exclusive use of the scientific and engineering research community—a global and growing community whose unbounded curiosity drives the breakthroughs that benefit humanity the most. From finding alternative energy sources to treating diseases to developing high-tech materials—the research done on leadership-class machines yields the world’s biggest technological advancements. These machines also serve to advance our understanding of our physical world, visible and invisible, the terrestrial and to galaxies far beyond our own.
Aurora will deliver more than eighteen times the computational performance of Mira, its predecessor at the ALCF, using a nearly equal number of compute nodes. Aurora will be a many-core system—but with nearly an order of magnitude more processors.
“The Leadership Computing Facility has been driving the evolution of supercomputing for over a decade. Its most important function is to align leadership systems with the needs and goals of breakthrough science projects. Aurora will deliver the system capabilities that our users require to expand their investigations in both scale and scope.”
— Michael E. Papka, Argonne Leadership Computing Facility
AURORA: HOW IT COMPARES
Aurora’s revolutionary architecture features Intel’s HPC scalable system framework and 2nd generation Intel® Omni-Path Fabric. The system will have a combined total of over 8 Petabytes of on package high bandwidth memory and persistent memory, connected and communicating via a high-performance system fabric to achieve landmark throughput. The nodes will be linked to a dedicated burst buffer and a high-performance parallel storage solution.
A second system, named Theta, will be delivered in 2016. Theta will be based on Intel’s second-generation Xeon Phi processor and will serve as an early production system for the ALCF.
System Features | Mira | Aurora |
---|---|---|
Compute Nodes | 49,152 | >50,000 |
Processor | PowerPC A2 1600 MHz | 3rd Generation Intel Xeon Phi |
System Memory | 768 TB | >7 PB DRAM and persistent memory |
System Interconnect | IBM 5D torus interconnect with VCSEL photonics | 2nd Generation Intel Omni-Path Architecture with silicon photonics |
File System Capacity | 26 PB GPFS | >150 PB Lustre |
Intel Architecture (x86-64) Compatibility | No | Yes |
Peak Power Consumption | 4.8 MW | 13 MW |
MADE FOR SCIENCE
- Climate Science
- Accelerator Design
- Biological Science
- Materials Science
- Transportation Efficiency
- Chemistry
- Cosmology
- Energy Storage
-
Climate Change Science
Understand dynamic ecological and chemical evolution of the climate system.
-
Accelerator Design
Explore and optimize innovative accelerator designs.
-
Biological Science
Enhance protein structure prediction and design capabilities.
-
Materials Science
Discover new materials for use in extreme conditions.
-
Transportation Efficiency
Improve transportation systems through design of enhanced aerodynamics features.
-
Chemistry
Investigate photoelectrochemical energy conversion for energy production.
-
Cosmology
Gain an understanding of the accelerated expansion of the Universe.
-
Energy Storage
Achieve breakthroughs in the development of electrode materials and electrolytes.
TIMELINE
2016
Open call for proposals to prepare scientific applications for Aurora and conduct pre-production science run campaigns.
Review proposals to select 10 Early Science Program projects.
Online/onsite training for ESP teams on Aurora hardware and programming.
On-site, hands-on workshops to prepare ESP users for Aurora
2016-2018
ESP teams modify, port, and optimize codes for Aurora.
2019
ESP teams get dedicated pre-production access to Aurora to conduct proposed science runs.
Target date for user access to Aurora
TRANSITIONING TO AURORA
Preparing Mira Users for Multi-Million-Core Computing
When Aurora is launched, it will be a state-of-the-art system for many-core processing. Its unprecedented capacity and revolutionary memory system will provide capabilities far beyond accelerated compute speed. Highly portable programming techniques already in use on present-day massively parallel systems, including Mira, apply directly to Aurora.
Aurora’s processing cores will run the same programs as mainstream Intel CPUs, and developers will enjoy a rich software ecosystem, including:
- MPI + OpenMP 4.x programming, which offers portability to commodity Linux clusters as well as CPU-GPU hybrid architectures.
- Intel compilers and optimization tools
- Cray compilers and libraries
- Third-party software including common analysis and visualization tools used in DOE supercomputing centers.
The ALCF’s Early Science Program will jump-start a set of large-scale scientific calculations. We will select ten projects based on an open call for proposals. These projects will port and optimize their application codes to Aurora’s architecture, and enjoy a period of pre-production Aurora access to establish the new frontier of computational science. En route, the ALCF will support these projects with:
- Training: tutorials and hands-on workshops
- A funded ALCF postdoctoral appointee for each project
- Access to early hardware: simulators, previous-generation Intel Phi systems, single nodes with Aurora’s processors, and, finally, mini versions of Aurora
- Technical assistance from the ALCF Intel-Cray Center of Excellence
The Early Science projects will share with the HPC community what they learned in preparing applications for Aurora via technical reports and community presentations.
As Aurora is built, the ALCF will organize and invite its user community to tutorials, webinars, and workshops to learn about Aurora’s architecture and how to program it. Users of Mira will enjoy a natural path to Aurora, starting with familiar programming models. Our goal is to illuminate that path so that all of our production users are ready to use Aurora’s transformative power from day one.
EARLY SCIENCE PROGRAM
Based on the highly successful ALCF Early Science Program for Mira, this program will prepare key applications for the architecture and scale of the new supercomputing resources, and solidify libraries and infrastructure to pave the way for other production applications to run on these systems. In addition to fostering application readiness for the future supercomputer, the Early Science Program allows researchers to pursue innovative computational science projects not possible on today’s leadership-class supercomputers.
Open to any scientist whose project has a demonstrable need for leadership-class resources, the ALCF will competitively award generous grants of pre-production time to run the scientific calculation that they have proposed. These projects should be representative of the spectrum of scientific areas and numerical methods found in production use of the ALCF. The PI-led project teams will consist of core developers of the application, domain science experts, and an ALCF postdoctoral appointee funded for the project. The teams will receive technical support from ALCF staff and the ALCF Intel-Cray Center of Excellence.
The ALCF will train the Early Science user community through tutorials and hands-on workshops. ALCF will share techniques used by the Early Science projects in preparing their applications for the new systems using technical reports and community presentations.