It’s time to meet the competitors at the SC20 Virtual Student Cluster Competition (VSCC). We’re going to introduce the teams, look at their records, and break down their games. We also have connected with our cluster competition bookie, Johnny Aces (pictured right) to see what he thinks of this year’s crop of competitors and get his predictions on the winners.

But first, a little bit about this year’s competition.

Like the preceding ISC20 competition, the SC20 event is completely virtual. This is, of course, due to the issues surrounding the COVID-19 virus. Where the ISC20 competition had students using the same cluster, the SC20 tourney is using Microsoft Azure as a hardware host for the student teams. Each team gets credits they can use on Azure to run their workloads on their selected configuration.

Teams will be measured by how much performance they can wring out of their Azure budget dollars as opposed to the power consumption metric used in years past.

As always, student teams will be competing to run a variety of HPC benchmarks and real-world applications to see which team can run the codes the fastest while making best use of their hardware resources.

The Teams

Team Clemson, U.S., aka “Team Dabo’s Tech Support”
Clemson is a first time cluster competition entrant. The team is named after the coach of their championship caliber football team (Go Tigers!), Dabo Swinney. They team has submitted applications to two previous SC competitions, but didn’t get over the hump until their third attempt this year. It’s difficult to evaluate a team when you haven’t seen them in a competition environment, but judging by the resumes of the team members, they look to be a strong newbie team on paper. The majority of them have held systems and HPC related internships and even full-time jobs. Odds to win: 10-1

Friedrich Alexander Universität, Germany, aka “Team default”
This is the thirteenth major cluster competition appearance for team FAU. They’ve made a name for themselves as HPL masters, having won the Highest LINPACK award at two previous events. Their highest overall finish was bronze, achieved recently at ISC20. They know their way around GPUs and optimizing code, but still haven’t put together the perfect game they need to take home a gold medal. But they’re a team on an upward trajectory, so this could be their year. Odds to win: 5-1

Tsinghua University, China, aka “Team Diablo”
This is the most experienced and most highly decorated team in student cluster competition history. Check out these stats: 20 competitions over the last eight years. ten gold medals, four silver medals, two bronze medals. Their teams are very stable over the years, meaning that each student has probably competed in several previous events. Team Tsinghua is definitely the team to beat in any competition. But with the Azure cloud-based competition, this is a new format for them, and they’ll need to adapt to it quickly. Odds to win: 3-1

ShanghaiTech University, China, aka “Team GeekPie_HPC”
This is the fourth competition for the team from Shanghai. They emerged in 2018 at the ASC event and surprisingly took home the silver medal. In their subsequent appearance at ISC18, they nailed HPCG and nabbed the Fan Favorite award. It will be interesting to see how the team performs this year. They have enough experience to know how the competition works and have tasted some success. Is this the year they drive into the top echelon? Odds to win: 5-1

Wake Forest University, U.S., aka “Team Daemon Deacons”
Third time competitor Wake Forest saw some success last year when they won the Reproducibility Challenge, which should give them some momentum going into the SC20 event. The team is returning two veterans, which should definitely help them in this year’s competition. They draw inspiration from their team anthem, Rick Astley’s “Never Gonna Give You Up,” proving that the 80s are alive and well in this generation and that some songs just won’t die! Odds to win: 10-1

University of Texas, Austin, U.S., aka “Team Longhorn”
This is the eighth SC competition for Team Longhorn. Over the years, they’re built up one of the best records in all of student clusterdom with three consecutive gold medals (2012-2014) and a Highest LINPACK award. They were also the first team to drive a mineral oil immersed cluster at SC11 in Seattle, which required an entire case of paper towels to maintain and manage. They haven’t competed in recent years, so this team is chock full of rookies. Can UT return to glory? Odds to win: 7-1

ETH Zurich, Switzerland, aka “Team RACKlette”

This team came out of nowhere at the ISC19 competition and nabbed the LINPACK Award and a bronze medal, instantly putting themselves on the radar of every cluster competition fan. They also appeared at SC19 and then again at ISC20. This team is sponsored by the Swiss National Supercomputing Center and coached by HPC luminary Torsten Hoefler. This is a smart, tough team that has been well trained. They’re great with their own hardware, but it’s an open question on how well they’ll adapt to the MS Azure cloud challenge. Johnny Aces, our in-house bookie, believes this team has a better than average chance of taking home the gold medal. Odds to win: 5-1

Texas A&M (plus San Antonio), U.S., aka “Team Aggie”
This is the third major international cluster competition for Team Aggie. They participated at SC18 in the US and also at ASC18 in China. This year, they’re reinforced by a group of students from the Texas A&M campus in San Antonio, Texas. It’s an entirely new team (including the coaches, I believe), so they’ll be learning about the competition as they go. It’s good to see the Aggies back in the competition, they have a special spirit. Odds to win: 10-1

University of California, San Diego, U.S., aka “Team Pink Flamingo”
This is another newbie team for SC20. UC San Diego, with assistance from the San Diego Supercomputing Center, is entering the competition with a student team that has a wide variety of experience with parallel systems, software development, and cloud computing. One interesting note is that UCSD does not currently offer an undergraduate HPC-centric computer science major. Maybe a big win at SC20 will change that, right? Odds to win: 9-1

University of Illinois at Urbana-Champaign, U.S., aka “Team UIUC”
This team is veteran heavy with three returning students from previous competitions, returning coaches, and new coaches who were past cluster competition team members. This will be their fifth major competition and the team has been steadily improving over the years. Their biggest sign of success was a bronze medal at the SC18 competition, but the team is hungry for more. Odds to win: 8-1

MIT, Boston University, U.S., aka “Team BUHPC”
This is the latest amalgamation team from Boston area universities, supported by the Massachusetts Green High Performance Computing Center. The Boston/MGHPC connection has run through thirteen previous teams, mostly powered by team coach Kurt Keville. They’ve competed at multiple events from China to Germany to SC events in the US. The team is known for their innovative hardware approaches in previous competitions – using Jetson GPUs, four-way nodes, and other interesting configurations. We’re looking forward to seeing what this latest team can do on MS Azure. Odds to win: 10-1

Nanyang Technological University, Singapore, aka “Team Supernova”
Nanyang has competed in twelve previous major competitions, having made appearances in all of the international tourneys. They’re not a team to be taken lightly, particularly when it comes to HPL as they’ve won three Highest LINPACK awards. Team Nanyang also took home the gold medal at the SC16 competition. They typically rely on low node count, high GPU clusters, so it will be interesting to see how they configure their Azure nodes. Odds to win: 4-1

Georgia Institute of Technology, U.S., “Team Phoenix”
We last saw Georgia Tech at the SC17 cluster competition, so it’s unlikely that many of the team members are vets of that team. Back then, they were a low node team, but also low GPU, which didn’t help their chances. I also think they had some problems with their hardware during the event, which also didn’t help much. However, they won’t have this problem at SC20, so it will give them a chance to strut their optimizing skills. Odds to win: 10-1

Northeastern University, U.S., aka “Team Husky”
The second entrant from the Boston area, Team Husky has competed at five previous competitions – all in the US. While previous teams have finished in the middle of the pack (so to speak…heh, heh, good one), they field solid teams as is befitting for a major research university. I could write another five or six dog-related puns here, but I’m not going to stoop to that level. Great to have you back, NEU, good luck this year. Odds to win: 10-1

North Carolina State University, U.S., “Team Wolf Pack”
This is the second consecutive SC competition for Team Wolf Pack. I thought the team was a bit handicapped by their two node, eight V100 cluster, and that they could have done a bit more with either more nodes or more GPUs. But the first year is typically the learning year for student cluster competition teams. It’s very rare to see first year teams rise above the middle of the pack (yeah, used the same pun again, ouch!). Time will tell if the experience they gained from SC19 will help the team improve their performance at SC20. However, the massively changed virtual competition this year will prove challenging to all returning teams. Odds to win: 10-1

Peking University, China, aka “Team Peking”
Team Peking has participated in four previous events, including SC17 and SC19. This school is a high-powered research institution, think “Harvard of China” caliber school. They would love to top their cross-town Beijing rivals, Tsinghua University, in this competition. In order to fuel this burgeoning rivalry, Peking will have to up their game. From meeting and talking to this team in past competitions, I’m fully convinced that they can and will do this. This may or may not be their year, but they can’t be overlooked. Odds to win: 9-1

Shanghai Jiao Tong University, China, aka “Team Shanghai”
This will be the tenth major competition for Team Shanghai. The school has bagged a gold medal at ASC14, two silvers, and a bronze, so they definitely have the skills to be a top echelon team. Moreover, their coach is a four-year cluster competition team captain, which gives the team a lot of experience from which to draw upon. They’ve been one of the most enjoyable teams for me to cover as well. I think this is a team that’s due to break out of the pack. Odds to win: 8-1

Southern University of Science & Technology, China, aka “Team SUST”
Team SUST is competing in their second cluster competition, coming off a successful rookie outing at ASC19. While the team didn’t win any major awards, they scored well on the applications, well enough to win an application optimization at the competition. We don’t know much about the SUST team coming into SC20 but based on our prior experience with them at ASC19, we can expect a solid team and a great effort from them. Odds to win: 10-1

University of Warsaw, Poland, aka “Team Warsaw”
Team Warsaw has quietly turned into a perennial competitor at major international cluster events. With eight competitions (this will be their nineth) in four years, they are rapidly building up the institutional knowledge and skills necessary to transform their team into a powerhouse. They’ve had some bumps along the way, mostly hardware related. What I like about this team is that they always bounce back no matter what happens, and that’s exactly what is needed in IT these days. Odds to win: 10-1

So there are your SC20 cluster competition teams. It’s an interesting mix of long-time competitors and complete newbies. Covering a virtual competition is difficult because we can’t just walk up to a team and ask them how it’s going, but we’ll do our best to get access to the teams and get as much news to you as possible. In the meantime, check out the dashboard, and watch the kick-off video, with comments from Student Cluster Competition Co-Chair Scott Michael and SC20 General Chair Christine E. Cuicchi.