Semiconductor & Computer Engineering
From WikiChip
Aurora - Supercomputers
| Edit Values | |
| Aurora | |
 | |
| General Info | |
| Sponsors | U.S. Department of Energy |
| Designers | Intel, Cray |
| Operators | Argonne National Laboratory |
| Introduction | 2023 |
| Peak FLOPS | 1,980 petaFLOPS |
| Price | $600,000,000 |
Aurora is a most powerful state-of-the-art exascale supercomputer designed by Intel/Cray for the U.S. Department of Energy (DoE) Argonne Leadership Computing Facility/Argonne National Laboratory (ALCF/ANL).
The system is expected to become the first supercomputer in the United States to break the exaFLOPS barrier.
Contents
Overview[edit]
With a price tag of around $600 million, Aurora is slated for 2023 and will reach a peak performance of over 1 exaFLOPS.
- Aurora (2023) - HPE Cray EX ° 2025/06 (3) ^ (TOP3!)
- • Intel Exascale Compute Blade,
- • Xeon CPU Max 9470 52C 2.4 GHz,
- • Intel Data Center GPU Max, Slingshot-11
- • Cores • Rmax (GFlop/s) • Rpeak (GFlop/s)
- • 9,264,128 • 1,012,000,000 • 1,980,006,000
History[edit]
Originally announced in April 2015, Aurora was planned to be delivered in 2018 and have a peak performance of 180 petaFLOPS. The system was expected to be the world's most powerful system at the time. The system was intended to be built by Cray based on Intel's 3rd generation Xeon Phi (Knights Hill microarchitecture).
In November 2017 Intel announced that Aurora has been shifted to 2021 and will be scaled up to 1 exaFLOPS. The system will likely become the first supercomputer in the United States to break the exaFLOPS barrier. As part of the announcement Knights Hill was canceled and instead be replaced by a "new platform and new microarchitecture specifically designed for exascale".
Original Specifications[edit]
| Original Specs | ||
|---|---|---|
| Model | BeBop / Theta (2017) | Aurora (2023) |
| Computing Power | ~180 (?) PFLOPS | 1,980 PFLOPS |
| Compute Nodes | >50,000 | |
| Processor | Intel Xeon Phi 7230 64C @1.3GHz, Cray CS400 3rd Generation Intel Xeon Phi (Knights Hill) |
Intel Xeon CPU Max 9470 52C @2.4GHz, Cray EX |
| Interconnect (Fabric) |
2nd Generation Intel Omni-Path Architecture with silicon photonics / Aries Interconnect |
Intel Exascale Compute Blade, Intel Data Center GPU Max, Slingshot-11 |
| Cores | 46,720 / 280,320 | 9,264,128 |
| Linpack Performance (Rmax) | 1.07 / 6.92 PFLOPS | 1,012 PFLOPS |
| Theoretical Peak (Rpeak) | 1.75 / 11.66 PFLOPS | 1,980 PFLOPS |
| Nmax | 3,408,706 / 7,680,000 | 28,773,888 |
| HPCG | 5,612 TFLOPS | |
| Memory | >7 PB DRAM and persistent memory | |
| File System | Intel Lustre* File System | |
| File System Capacity | >150 Petabytes | |
| File System Throughput | >1 Terabyte/s | |
| Peak Power | 13,000 kW (?) / 1,087 kW | 38,700 kW |
| FLOP/s Per Watt | >13 GFLOP/s per watt | |
| Facility Area | ~3,000 sq. ft. | |
Models[edit]
- Supercomputers • DoE • ANL
- Aurora • 2023 ° 2025/06 (3) ^ (TOP3!)
- Improv • 2023
- Polaris • 2021
- Theta • 2017
- BeBop • 2017
- Cooley • 2015
- Mira • 2012
- Cetus • 2012
- Vesta • 2012
- Blues • 2012
- Magellan • 2010
- Jazz 2 • 2009
- Intrepid • 2007
- eServer • 2005
- Jazz • 2002
TOP500[edit]
DOE/SC/Argonne National Laboratory URL: http://www.anl.gov/ Segment Research City: Argonne Country/Region: United States System Year Vendor • Cores • Rmax (GFlop/s) • Rpeak (GFlop/s) *Aurora - HPE Cray EX - Intel Exascale Compute Blade, Xeon CPU Max 9470 52C 2.4 GHz, :Intel Data Center GPU Max, Slingshot-11 ° 2025/06 (3) ^ :2023 Intel • 9,264,128 • 1,012,000,000 • 1,980,006,000 *Improv - PowerEdge R6525, AMD EPYC 7713 64C 2.0 GHz, Infiniband :2023 DELL • 105,600 • 2,509,700 • 3,379,200 *Polaris - Apollo 6500, AMD EPYC 7532 32C 2.4 GHz, NVIDIA A100 SXM4 40 GB, Slingshot-10 :2021 HPE • 256,592 • 25,810,000 • 34,163,190 *Theta - Cray XC40, Intel Xeon Phi 7230 64C 1.3 GHz, Aries interconnect :2017 Cray/HPE • 280,320 • 6,920,900 • 11,661,312 *BeBop - Cray CS400, Intel Xeon Phi 7230 64C 1.3 GHz/Xeon E5-2695v4, Intel Omni-Path :2017 Cray/HPE • 46,720 • 1,070,590 • 1,750,016 *Cooley - Cray CS300-AC, Xeon E5-2620v3 6C 2.4 GHz, Infiniband FDR, Nvidia K80 :2015 Cray/HPE • 4,788 • 240,400 • 293,722 *Mira - BlueGene/Q, Power BQC 16C 1.60 GHz, Custom :2012 IBM • 786,432 • 8,586,612 • 10,066,330 *Cetus - BlueGene/Q, Power BQC 16C 1.60 GHz, Custom Interconnect :2012 IBM • 65,536 • 715,551 • 838,861 *Vesta - BlueGene/Q, Power BQC 16C 1.60 GHz, Custom :2012 IBM • 32,768 • 357,776 • 419,430.4 *Blues - Appro Xtreme-X Supercomputer, Xeon E5-2670 8C 2.60 GHz, Infiniband QDR :2012 Cray/HPE • 5,184 • 81,820 • 107,827 *Magellan - iDataPlex, Xeon X55xx QC 2.66 GHz, Infiniband :2010 IBM • 4,032 • 38,659.3 • 42,997.3 *Jazz 2 - iDataPlex, Xeon E55xx QC 2.53 GHz, Infiniband :2009 IBM • 2,640 • 22,333.9 • 26,716.8 *Blue Gene/P Solution :2008 IBM • 4,096 • 11,710 • 13,926.4 *Intrepid - Blue Gene/P Solution :2007 IBM • 163,840 • 458,611 • 557,056 *eServer - Blue Gene Solution :2005 IBM • 2,048 • 4,713 • 5,734 *Jazz - LCRC Xeon 2.4 GHz - Myrinet :2002 Linux Networx • 361 • 1,007 • 1,732.8
See also[edit]
Documents[edit]
Original System[edit]
- Ushering in a New Era, Argonne National Laboratory’s Aurora System, April 2015
- Aurora Fact Sheet
External links[edit]
References[edit]
- Introducing Aurora, April 9, 2015
- U.S. Department of Energy Selects Intel to Deliver Nation’s Most Powerful Supercomputer at Argonne National Laboratory, April 9, 2015
- Unleashing High-Performance Computing Today and Tomorrow, November 13, 2017
Facts about "Aurora - Supercomputers"
| designer | Intel + and Cray + |
| introductory date | 2023 + |
| main image |  + |
| name | Aurora + |
| operator | Argonne Leadership Computing Facility + and Argonne National Laboratory + |
| peak flops (double-precision) | 1.98e+18 FLOPS (1.98e+15 KFLOPS, 1,980,000,000,000 MFLOPS, 1,980,000,000 GFLOPS, 1,980,000 TFLOPS, 1,980 PFLOPS, 1.98 EFLOPS, 0.00198 ZFLOPS) + |
| release price | $ 600,000,000.00 (€ 540,000,000.00, £ 486,000,000.00, ¥ 61,998,000,000.00) + |
| sponsor | U.S. Department of Energy (DoE) + |
