News and Announcements
TOP500: June 2020
Photo of Fugaku Supercomputer courtesy of RIKEN-CCS
The 55th TOP500 list was just unveiled at this year’s ISC High Performance Computing (ISC-HPC) conference on June 22, 2020. The new Fujitsu Fugaku system, installed at the RIKEN Center for Computational Science (RIKEN-CCS) took the no.1 spot on the TOP500 with a stunning 415.5 PFLOP/s on the High Performance Linpack (HPL) benchmark—handily topping the Summit machine in Oak Ridge. The Fugaku system is powered by Fujitsu’s 48-core A64FX SoC, becoming the first no. 1 system on the TOP500 to be powered by ARM processors.
The rest of the TOP5 machines are veterans of the list, but two new systems edged their way between Tianhe-2A (National University of Defense Technology, China) and Frontera (Texas Advanced Computing Center, USA).
Now occupying no. 6 is the HPC5 system, built for the Italian energy firm Eni S.p.A, which achieved 35.5 PFLOP/s in HPL, making it the fastest supercomputer in Europe. HPC5 is a Dell PowerEdge system boasting Intel Xeon Gold processors and NVIDIA Tesla V100 GPUs over a Mellanox HDR InfiniBand network.
Selene now occupies no. 7 with 27.58 PFLOP/s in HPL. Selene, installed at NVIDIA’s facilities in the United States, is based on the new DGX SuperPOD and is powered by NVIDIA’s new Ampere A100 GPUs and AMD’s EPYC Rome CPUs.
Click here to see how the rest of the TOP500 panned out.
| Rank | System | Cores | Rmax (TFLOP/s) | Rpeak (TFLOP/s) | Power (kW) |
|---|---|---|---|---|---|
| 1 | Supercomputer Fugaku – Supercomputer Fugaku, A64FX 48C 2.2GHz, Tofu interconnect D, Fujitsu, RIKEN Center for Computational Science, Japan |
7,299,072 | 415,530.0 | 513,854.7 | 28,335 |
| 2 | Summit – IBM Power System AC922, IBM POWER9 22C 3.07GHz, NVIDIA Volta GV100, Dual-rail Mellanox EDR Infiniband, IBM, DOE/SC/Oak Ridge National Laboratory, United States |
2,414,592 | 148,600.0 | 200,794.9 | 10,096 |
| 3 | Sierra – IBM Power System AC922, IBM POWER9 22C 3.1GHz, NVIDIA Volta GV100, Dual-rail Mellanox EDR Infiniband, IBM/NVIDIA/Mellanox, DOE/NNSA/LLNL, United States |
1,572,480 | 94,640.0 | 125,712.0 | 7,438 |
| 4 | Sunway TaihuLight – Sunway MPP, Sunway SW26010 260C 1.45GHz, Sunway, NRCPC, National Supercomputing Center in Wuxi, China |
10,649,600 | 93,014.6 | 125,435.9 | 15,371 |
| 5 | Tianhe-2A – TH-IVB-FEP Cluster, Intel Xeon E5-2692v2 12C 2.2GHz, TH Express-2, Matrix-2000, NUDT National Super Computer Center in Guangzhou, China |
4,981,760 | 61,444.5 | 100,678.7 | 18,482 |
Report on the Fugaku Supercomputer
On June 22, 2020, Jack Dongarra released his report on RIKEN’s new Fugaku Supercomputer. As outlined in the report and discussed in the video above, the new Fugaku system hit 415.5 PFLOP/s on the HPL benchmark, besting the now second-place Summit system by a factor of 2.8×.
Featuring 7,299,072 cores worth of Fujitsu’s 48-core A64FX SoC—and 4,866,048 GB of total memory spread across 156,976 nodes—Fugaku is also the first no. 1 system on the TOP500 to be powered by ARM processors. Bucking another trend, Fugaku does not use GPU accelerators.
For more details, read the full Fugaku Report here.
HPCG: June 2020
The latest results for the HPC Preconditioned Conjugate Gradient (HPCG) benchmark were also released at ISC-HPC 2020. A joint effort between ICL and Sandia National Laboratories, HPCG is designed to measure performance that is representative of modern HPC capability by simulating compute and communication patterns from sparse iterative solvers commonly found in science and engineering applications.
HPCG results are released twice per year alongside the TOP500 rankings to show how real-world applications might fare on a given machine. Here, as with the HPL-based TOP500, RIKEN’s Fugaku system takes the top spot from Summit with 13.4 PFLOP/s in HPCG. HPC5 also cracks the top 5 systems with 0.860 PFLOP/s in HPCG. The full list of HPCG rankings is available here.
| Rank | Computer | HPL (PFLOP/s) | TOP500 Rank | HPCG (PFLOP/s) | %Peak |
|---|---|---|---|---|---|
| 1 | Supercomputer Fugaku – Fujitsu, A64FX 48C 2.2GHz, Tofu interconnect D
RIKEN R-CCS, Japan |
415.5 | 1 | 13.4 | 2.6% |
| 2 | Summit – IBM, POWER9, NVIDIA Volta V100
DOE/SC/ORNL, USA |
148.6 | 2 | 2.926 | 1.5% |
| 3 | Sierra – IBM, Power9, NVIDIA Tesla V100
DOE/NNSA/LLNL, USA |
94.64 | 3 | 1.796 | 1.4% |
| 4 | HPC5 – Dell PowerEdge, Intel Xeon-Gold, NVIDIA Tesla V100
Eni Green Data Center, Italy |
35.45 | 6 | 0.860 | 1.7% |
| 5 | Trinity – Cray XC40, Intel Xeon E5-2698 v3, Xeon Phi 7250
DOE/NNSA/LANL/SNL, USA |
20.159 | 11 | 0.546 | 1.3% |
Holberton Puerto Rico
ICL alumnus Adam Beguelin is opening up a coding school in Puerto Rico! Holberton Puerto Rico, a branch of Holberton San Francisco, is a project-based alternative to college, where there is no up-front tuition and no prior coding experience needed.
You can email Adam (adam@codepuertorico.com) if you have any questions, and Holberton is now taking applications for the September cohort!
Employment Opportunities at ICL
ICL is seeking full-time Research Scientists (MS or PhD) to participate in the design, development, and maintenance of numerical software libraries for solving linear algebra problems on large, distributed-memory machines with multi-core processors, hardware accelerators, and performance monitoring capabilities for new and advanced hardware and software technologies.
The prospective researcher will coauthor papers to document research findings, present the team’s work at conferences and workshops, and help lead students and other team members in their research endeavors in ongoing and future projects. Given the nature of the work, there will be opportunities for publication, travel, and high-profile professional networking and collaboration across academia, labs, and industry.
An MS or PhD in computer science, computational sciences, or math is preferred. Background in at least one of the following areas is also preferred: numerical linear algebra, HPC, performance monitoring, machine learning, or data analytics.
For more information check out ICL’s jobs page: http://www.icl.utk.edu/jobs.
Conference Reports
ISC High Performance 2020
The 2020 ISC High Performance Computing (ISC-HPC) conference kicked off on June 22, 2020 all over the world for the conference’s first “digital event.” The initial opening session and TOP500 were live streamed, and the rest of the paper presentations and focus sessions are available for those who registered to attend.
Jack Dongarra was “Zoomed” in to the TOP500 and TOP500 Q&A focus sessions, where Erich Strohmaier “handed” out the certificates for this summer’s TOP500, Green500, and HPCG awards. Jack, Erich, and Satoshi Matsuoka (Director of RIKEN-CCS) then exchanged in a Q&A with Horst Simon (on behalf of the audience).
ICL collaborators from KIT, including Hartwig Anzt, Mike Tsai II, and Pratik Nayak also joined the ISC-HPC virtual ranks. For his part, Mike Tsai II presented work on “Sparse Linear Algebra on AMD and NVIDIA GPUs.” According to Hartwig, this topic garnered significant interest, including coverage in The Next Platform and direct inquiries from AMD and NVIDIA. Pratik also presented work on “Evaluating Asynchronous Schwarz Solvers for GPUs” to mitigate problems with synchronizing resources across multiple nodes by using an asynchronous Restricted Additive Schwarz method.
And while being grounded with no travel has been frustrating and even maddening for some, it has unquestionably opened up events like ISC-HPC in unprecedented ways.
The Editor would like to thank Hartwig Anzt for his contributions to this article.
Interview
Yicheng Li
Where are you from, originally?
I am from Shanghai, China.
Can you summarize your educational background?
I did K–9 in China, but I actually went to high school (grades 10–12) in Memphis, TN. I earned my BS from UTK in 2018, and I have continued on at UTK for my PhD.
Where did you work before joining ICL?
I worked with Dr. Yilu Liu’s group as an Undergraduate Research Assistant. While in the group, I worked on the MobileUGA (Universal Grid Analyzer) app, which is intended to monitor the power grid by collecting data from outlets.
How did you first hear about the lab, and what made you want to work here?
My (now) fellow ICLers Qinglei and Yu told me about ICL. It’s in the comfort zone to stay in school.
What is your focus here at ICL? What are you working on?
I’m working on my PhD and concentrating on an MPI datatype engine. We’re looking to maximize performance on pack/unpack operations by exploiting new ways of describing data layouts.
What are your interests/hobbies outside of work?
I enjoy playing basketball.
Tell us something about yourself that might surprise people.
I’m a jock.
If you weren’t working at ICL, where would you like to be working and why?
Not sure yet. Still finding something that will interest me (besides research).
