The student supercomputing team from Tsinghua University showcased exceptional computing skills to win the overall championship at the ISC 2024 Student Cluster Competition. The finals, which concluded in the afternoon of May 15 in Hamburg, Germany, saw Tsinghua’s team also achieve the Highest LINPACK Benchmark Score award.
The victory marks Tsinghua University’s 17th win across the three major international student supercomputing competitions, solidifying its reputation in the field. Notably, this is the 7th time Tsinghua has emerged victorious since the inception of the ISC Student Cluster Competition in 2012.
Under the constraints of a 6-kilowatt power limit, each team was tasked with constructing their own server cluster. The rigorous competition tested their systems across various representative benchmarks (including LINPACK, HPCG, HPCC) and scientific computing applications such as Neko, Conquest, and RegCM. A mystery application, the lattice Boltzmann method simulation OpenLB, was also part of the challenge.
Affected by the pandemic, this year’s competition marked the first in-person participation for the team since 2019. The members dedicated their time since winter vacation to preparation, honing their skills for the event. Their efforts paid off as they navigated through jet lag, power consumption limit, and technical challenges to secure the top spot with a comprehensive skill set and stable performance.
In a standout achievement, Tsinghua University achieved first place in the LINPACK benchmark with an impressive score of 337 TFLOPS.
The ISC24 Student Cluster Competition, hosted by the HPC-AI Advisory Council, is one of the three most prestigious international student supercomputing contests, along with the ASC Student Supercomputer Challenge and the SC Student Cluster Competition. This year’s live final featured eight teams from seven countries. The winning Tsinghua team comprised six undergraduates: Runqing Zhang, Jingbo Shan, Xingye Yuan, and Kai Yang from Zhili College, Zhiyu Xue from the Department of Computer Science and Technology, and Jiaqi Pan from the Institute for Interdisciplinary Information Sciences. The instructors are lecturer Wentao Han, professor Jidong Zhai, and postdoctoral fellow Yuyang Jin of the Department of Computer Science and Technology. Ph.D. students Shengqi Chen and Mingshu Zhai from the Institute of High Performance Computing of the Department of Computer Science and Technology provided technical support for the training.
