On February 1, 2023, People's Daily published a report in the column "Promoting the Spirit of Scientists", telling the stories of the teachers and students from the Graph Computing Team of the School of Computer Science and Technology, Huazhong University of Science and Technology (HUST). The report records how they have been working to tap deeper into the research of graph computing research, breaking and keeping world records during the past few years.
Above: Yu Zhang(second from right) leads JinZhao (second from left) and other team members to do research in the laboratory.
Below: Hai Jin participates in an academic forum.
(Photos Provides by HUST)
The graph computing team of the School of Computer Science and Technology of HUST has been landing new world records in graph computing in recent years, winning a myriad of international awards in this region. Founded after Professor Hai Jin's return in 2000, the laboratory has undertaken a number of national key scientific research projects. His team has won the Second Prize of the National Natural Science Award once, the Second Prize of the National Scientific and Technological Progress Award twice, and the Second Prize of the National Technology Invention Award once, boasting dozens of international invention patents.
In November 2022, among the top 500 graph computing engines listed at the 25th International Supercomputing Conference (ISC), based on "Pengcheng CloudBrain II", the engine developed by the graph computing team of HUST took the lead in terms of single-source shortest path performance, and tops among domestic processors in terms of breadth-first search performance.
As members of the R&D team in charge of the project, Yu Zhang, an associate professor at the School of Computer Science and Technology, HUST, and Jin Zhao, a doctoral student, are very excited. "Despite we are a team with an average age of 25, we made breakthroughs through collaboration and finally tackled thorny world-class problems," said Zhang.
From zero to international awards
At present, global big data has entered a period of accelerated development. As the key technology of next-generation artificial intelligence, graph computing has been widely applied in many fields such as medical care, education, and finance. The Graph500 is an authoritative list for evaluating the graph computing performance of supercomputers jointly organized by the ISC and the Institute of Electrical and Electronics Engineers (IEEE). Graph processing is an important big data algorithm model. The graph here does not refer to an "image", but an abstract data structure. Based on this, researchers need to conduct individual analysis, relationship matching, and complex calculations to realize knowledge reasoning, event tracing, and prediction.
Today, many young science junkies have become fans of HUST's team. "Putting our name on the list is not our goal. Based on our research in graph processing since 2010, we will move on to industrial applications of our achievements to better serve the country's economic and social development," said Zhang.
When the concept of graph computing was first proposed, Zhang had a strong interest in it: "At that time, I realized this would be a hot-button research direction in the future." In 2011, Zhang began his Ph. D with high-performance graph processing system software and architecture, starting his long journey of scientific research.
In recent years, this young team has won a myriad of awards in international competitions and made new world records. On November 18, 2021, in the rankings of the 18th Green Graph500 and the 23rd Graph500 announced at the ISC, they won two awards for leading performance-to-power ratio and stand-alone performance. In 2022, in the 24th Graph500, the performance of their graph computer once again break the record, leading the performance of a single machine to surpass the graph computing performance of a supercomputer for the first time.
"The performance-to-power ratio tops means our graph computer can process more edges per watt than other competitors. The stand-alone performance means that our graph computer has the highest edge processing speed among all the single machines. Thousands of similar machines are not as fast as one of ours," he said.
Zhang told People's Daily that these achievements are inseparable from the support of the laboratory and Professor Jin's efforts. Jin is the Director of the National and Local Joint Engineering Research Center for Big Data Technology and Systems at HUST, the Director of the Key Laboratory of Service Computing Technology and Systems of the Ministry of Education, and the Hubei Provincial Key Laboratory of Cluster and Grid Computing. In 2000, he completed his postdoctoral research abroad and returned to China to establish this laboratory. At that time, China's research in related fields had just started, and the laboratory started with purchasing equipment, finding a site, and setting up a project. When recalling all the hardships, Jin still feels impressed, "In recent years, the laboratory has ranked first on the lists many times, attracting the attention of the industry. As a matter of fact, we have been working on relevant research for over ten years. This is a result of long-term accumulation," said Jin.
Aim at the academic frontier to satisfy the country's major strategic needs
As early as 2015, Jin discovered the huge application value of graph computing: "Graphs are the most common carrier to express the relationship between things in the era of big data. The ability to process big data is the ability to process graphs." Therefore, the laboratory proposed a plan to use a tailored acceleration chip for graph processing. However, since the graph computing-related technology and the software system were still in the process of development, many peer experts were not bullish on this plan and took it as impractical, and some even raised objections. But, based on his own research and judgment, Jin persisted and started the research.
Later, during the "13th Five-Year Plan" period, China approved a key research and development plan "General Computer Technology and System Oriented to Graph Computing". This plan listed as excellent in the semi and final evaluation is led by Professor Xiaofei Liao from the laboratory.
Jin's judgment got approved. Today, his laboratory has become a leading scientific research team with more than 40 teachers and over 300 students, expanding the research from parallel computing and high-performance computing to distributed computing, cloud computing, and Big Data.
"Difficulties arise at any time, and we always feel pressure!" Jin Zhao, a post-95 doctoral student, admitted the hardships in scientific research. "The team members are stuck in the laboratory almost all year round. The teacher instructed us to do research all day long." Once, Zhao encountered a bottleneck in terms of performance while optimizing data access for graphing computing and still had no breakthrough despite several months of deliberation and research. When he saw his peers with many published papers, he even had the idea of giving up. But his teacher Zhang has always encouraged him: "Be brave to tackle world-class difficult problems, and don't be satisfied with doing simple things." In the end, under the guidance of Professor Zhang, Zhao successfully solved the problem together with other graph computing experts, and presented the achievements at international conferences.
Although Qiange Shen is a newcomer to this team, she has already felt the challenge of doing scientific research. Once, the code of some tasks she oversaw went wrong and was not corrected after checking a lot of information. "It was a complicated program to be solved within a short time, therefore the pressure makes me unable to fall asleep. Eventually, I found out all the problems after staying up several nights. This is my only experience of such difficulty, but I believe the professors and other seniors must have had more."
It is precisely the endeavor over the past ten years that has built the team's current first-class computing speed. "Now it seems right to come back to China. To do scientific research, we should meet the country's major strategic needs," said Jin.
Leniency and strictness hand down a fine learning spirit
One rule in the laboratory team is to keep the word "strict" in mind. Professor Zhang is Zhao's Ph.D. paper supervisor. "The professor had several scientific research projects at that time. Time being so limited, he still helped revise my paper word by word, marked every modification, and explained them to me in case of making the same mistake in the future." It took more than a month to correct a paper, but Zhang Yu didn't find it hard. "Professor Jin revised my paper like this when I was a Ph. D student. Now I have my own students and can understand what a teacher expects. The strict requirements now are for them to be more competitive and resilient in the future."
Although the work in the laboratory is hard, there are also some unforgettable warm moments. Zheng Long, a student in the lab, recalled receiving a blessing from Professor Jin on his birthday during the internship in his junior year. "I was very moved when I received the blessing, and I made up my mind to make some achievements." He receives Professor Jin's blessing every year since then. Every member of the lab receives the professor's birthday blessing.
In the corridor of the laboratory, several large exhibition boards display the various honors and awards the team has received since its establishment, its invention patents, the group photos of the team's activities, and a photo wall composed of the smiling faces of all the teachers and students in the laboratory. This is inspiring and heartwarming. A few lines on the display board catch visitors' eyes: "A teacher cannot educate others unless he is still studying tirelessly; a lamp can never light another lamp unless it continues to burn its own flame." "This passage touched me very much. Since I returned to China to become a teacher, I have been hanging it on the wall of the lab as my motto. Every new teacher who enters the laboratory will receive a nameplate with these words on it." Since the establishment of the laboratory more than 20 years ago, Jin and his team have been marching forward on the road of scientific research to serve the country.
Struggle and hard work build the stone steps toward the peak
During the interview, what impressed the reporter most was the team members' spirit of rising to the challenge and maintaining unremitting struggle. In the very beginning, even when their research direction was questioned or even opposed, the team members did not give up their persistence in their ideals and their commitment to scientific research, moving from 0 to 1. Now, they still choose to challenge the world-class difficult problems, face up to difficulties, and pursue excellence.
Sharpening makes a mighty sword. The courage to challenge comes from scientific research capabilities, teamwork, and the days and nights they stick to the laboratory. "I arrive at the laboratory at 8 am every day and leave at 10 pm. Sometimes when I return to the dormitory, I still think about the problems unsolved during the day." "As long as you don't think about scientific research problems before going to bed, you won't suffer from insomnia. On some days, I'm still dealing with program bugs at two o'clock in the morning." Indeed, it is every day and night of persistence and hard work that builds the stone steps toward the peak.
Promote the spirit of scientists
Stay committed to making scientific and technological innovations
The School of Computer Science and Technology
Will be more resolved to make continuous efforts
In the year 2023!