On February 4th, good news arrived from the 2024 ASC Student Supercomputer Challenge (ASC24), one of the top three global supercomputing events alongside the United States’ SC and Germany’s ISC competitions. Three undergraduate teams from the School of Computer Science performed excellently, securing three second-place awards after intense competition.
ASC24 Team One. Photo Provided by the School of Computer Science
ASC24 Team Two. Photo Provided by School of Computer Science
ASC24 Team Three. Photo Provided by School of Computer Science
The preliminary round’s topics covered high-performance computing (HPC) system design, performance optimization for HPL and HPCG, large model inference optimization, and percolation numerical simulation. The latter two focused on artificial intelligence and real engineering challenges, with the large language model inference optimization requiring teams to build and optimize an inference engine based on the popular open-source model LLaMA2, aiming for the highest possible throughput on a dataset of 10,000 samples provided by the organizers. The percolation numerical simulation aimed to study the complex flow of multiphase fluids in porous media, simulating the permeation of oil, gas, and water in oil extraction scenarios and optimizing large-scale parallel computing processes to improve the computational performance and efficiency of discrete algorithms.
These tasks span multiple domains of supercomputing and artificial intelligence, testing the teams’ analytical skills in supercomputing applications, performance optimization, and parallel strategy design. Our three participating teams, composed of 14 undergraduates from three grades and four majors—Artificial Intelligence, Computer Science and Technology, Network Engineering, and Software Engineering—engaged actively in the competition. Despite challenges like overlapping competition and final exam schedules and using the computing power provided by the Wuhan Artificial Intelligence Computing Center for the first time, they successfully submitted their technical reports in English. These reports detailed cluster design, source code optimization methods, and output files, demonstrating impressive technical strength, innovation, and comprehensive quality, as well as teamwork.
Members from SCMU Participating in ASC24. Photo Provided by School of Computer Science
The ASC Student Supercomputer Challenge aims to cultivate young talent, promote the research, development, and innovation of supercomputers, and encourage their widespread application across various industries. Since its inception in 2012, the competition has grown in scale from China to Asia and then globally, becoming the world's largest and most participated university supercomputing event. In this year’s competition, 89 teams, including those from Beijing University of Technology, Beijing Normal University, Tsinghua University, Hong Kong University of Science and Technology, Zhejiang University, Wuhan University, and Fuzhou University, were awarded second prizes.
The School of Computer Science places great emphasis on cultivating students’ supercomputing application skills. Led by the Department of Artificial Intelligence and established at the end of 2018, the supercomputing team is guided by professional teachers including Tang Wan, Ye Zheng, Zhang Xiao, Yang Ximin, and Cheng Peng, with more than 20 members currently. To date, the school has organized students to participate in four ASC Student Supercomputer Challenges, with all 10 teams receiving second-place awards. We look forward to more students joining and exploring the limitless possibilities in the field of supercomputing.
Edited by Long Hua, Reviewed by Lei Changsheng
