Loading...

Founding Chair of BenchCouncil

Full Professor, University of Chinese Academy of Sciences & ICT, Chinese Academy of Sciences

Office: Room 1038, NO. 6, Kexueyuan South Road, Zhongguancun, Haidian District, Beijing, PRC

Email: jianfengzhan.benchcouncil@gmail.com

About Me (PDF)

A common thread among my research projects focuses on benchmarking, designing, implementing, and optimizing parallel and distributed systems. I have made substantial and effective efforts to transfer my academic research into advanced technology to impact general-purpose production systems. Several technical innovations and research results from my team have been widely adopted in benchmarks, operating systems, and cluster and cloud system software, with direct contributions to advancing parallel and distributed systems in China and even the world.

My statement of research interests, objectives, and vision

In the future five years (2024-2029), I plan to work on the following research topics:

  • Benchmarkology ( :-) fortunately, I coined this term). In all disciplines, how to evaluate is a fundamental issue. Unfortunately, most evaluation approaches are ad-hoc. Proposing a unified evaluation principle and methodology is a very challenging issue. Recently, I have had some intuitions about the solution to this challenge. Metrology is the measurement science and application which objectively observes or measures objects or things. Instead, evaluation is subjective. I propose a unified benchmark-based evaluation approach. I have been building the underlying mathematic model, principle, and methodology, which I call benchmarkology. As a case study, I am working on how to evaluate intelligence, chips, and fundamental algorithms with my Ph.D. students and colleagues.

  • Benchmarkology-guided exploration of the solution instantiation space. In my previous work, I propose that a benchmark is an explicit or implicit definition of a problem. Based on the benchmarkology, I will revisit the computer system issues from the top down, beginning with systematically defining the problems and then exploring the solution instantiation space. I will examine the possibility of proposing new computations or management abstractions.

  • Opensource computer systems initiative (OpenCS). As a wild idea, I am defining the benchmarks of global challenges. I launched an open-source computer system initiative to tackle those challenges. It will not reinvent the wheel. Instead, I will propose new abstractions and methodologies to tackle these fundamental challenges.

Statement of Teaching Interests & Philosophy

I am not a traditional teacher. I ever taught courses. But I have not taught a lesson in the past ten years. My experience is supervising Ph. D students and Master students through investigating challenging projects.

I ever draft an article to propose three laws to govern the technology rise or fall. Before reading the following summary, please refer to my previous article, "Three Laws of technology rises or Falls."

I summarize my teaching interests and philosophy in four points.

  • Studying through investigating a challenging project is the best learning process or experience.

  • If you fail to achieve on a challenging issue, I suggest you work on an open-source tool valuable to the community. There are three reasons to justify it. First, most junior researchers are not qualified to tackle a significant challenge. Second, a tool valuable to our community will gain your reputation. Three, working on a tool will train your skill which will help you find a job.

  • If you are experienced and confident, I encourage you to find and tackle a challenging issue. Even solving a challenge in ten years is worthwhile. But you cannot expect too much.

  • System innovation is very challenging and needs a tradeoff. Many guys plan to build innovative system prototypes. But as a Ph. D student or a junior researcher, you must consider people and time budget. With a limited budget, the systems are error-prone or too simple to justify their motivation. The first importance of all, you need to define a minimum set of functionalities to showcase the convincing innovation and demonstrate the change force. After the change force is validated, it is time to implement a fledged function with a double budget.