Award
2020 NSF CAREER Award Winner
Department
Department of Computer Science
Awarded Project
Steve Jian’s MemMax project explores how to co-design CPU and OS to maximize memory utilization in cyberinfrastructure systems, to boost their performance growth. MemMax consists of two research thrusts, one targeting HPC systems and one targeting cloud systems, as these two types of systems have different causes for their memory underutilization, states Jian. His research methodology consists of real-system measurements to characterize the behavior of existing systems, hardware prototyping to valid the functional correctness of MemMax, and architectural simulations to quantify the performance improvement MemMax achieves.
What path did you take to get to this point in your career and research?
Ever since being introduced to my first computer — an old computer running Windows 3.1 — as a young kid, I have been amazed by how rapidly computers have been improving. Throughout my childhood and teenage years, after every one or two years, one could buy a new computer that is noticeably faster than and at the same price as before. Such exponential growth in performance/dollar (i.e., the amount of computation one could buy for the same cost) has radically transformed our world by making computing ubiquitous, not only on our desks, but also in our phones, watches, cars, fridges etc.
Unfortunately, when I entered graduate school, it became clear that sustaining this exponential improvement has become increasingly difficult as cost-effectively making transistors (i.e., computers’ building blocks) faster is very difficult as transistors were already very fast. This motivated me to pursue research in the field of computer architecture, which seeks to speed up computers by exploring better hardware algorithms, as opposed to making transistors faster.
What impact do you hope your research will have?
My current research is to explore new computer system designs to improve the performance/dollar for large-scale computing systems such as supercomputers and cloud. Improving the performance/dollar of supercomputers will accelerate scientific and engineering discoveries spanning weather prediction, health and medicine, transportation, and energy production. Improving performance/dollar for cloud will help build better economies by enabling new digital products/services that require faster and cheaper computing.
Specifically, my current work is improving the memory systems of these large-scale systems. In supercomputers, memory performance significantly impacts overall system performance, as supercomputing workloads are typically memory-intensive due to processing large volumes of data. Here, I am working on redesigning the memory system to provide lower latency and higher bandwidth. In cloud, memory capacity largely determines pricing for users as each cloud server typically runs hundreds of user applications, which have to contend for a limited pool of memory. I am working on redesigning the memory system to pack as much information as possible in the same amount of physical memory while preserving high performance.
What do you find most interesting about your field of engineering?
Computer science is a dynamic field. It is often said that a decade in computer science is a century in other sciences. New computer hardware technologies constantly emerge. This enables new computer software that in turn motivates the design of new computer hardware. Such a dynamic field provides many new problems to explore and solve, which is exciting for researchers.
If you had one piece of advice to give students that aspire to pursue research and are just starting their journey, what would you share with them?
Research may seem like very hard work at times. During the hard times, stay encouraged by remembering that few things in life can beat the excitement of being forever known by the world as the first person/group of people to have solved an important problem when your work finally gets published.