Past Events

Faculty Candidate Talk Emerging Architectures for Humanity's Grand Challenges
× Export Event Save iCal Send to Google Calendar Send to Yahoo Calendar Send to Outlook Live Send to Microsoft Outlook Save iCal Preserve formatting in description (only supported in some calendar applications) Close Tuesday, February 25, 2020, 10:30am

Speaker: Yipeng Huang, Princeton University

Bio

Yipeng is a postdoctoral research associate at Princeton University, working with Prof. Margaret Martonosi. He received his PhD in computer science in 2018 from Columbia University, working with Prof. Simha Sethumadhavan. His research interest is in building, programming, and in identifying applications for emerging architectures. These include quantum and analog computer architectures that may uniquely address challenges in scientific computing, but would require new programming tools and architectural abstractions. His work has been recognized as an IEEE Micro Top Pick among computer architecture conference papers. He has also received support for his research work through DARPA, in the form of a Small Business Technology Transfer grant to investigate commercial applications of his research.

Location : CoRE A 301

: 

Event Type: Faculty Candidate Talk

Abstract: As we enter the post-Moore's law era of computer architecture, researchers are turning to new models of computation to address humanity's Grand Challenges. These new models of computation include analog computing and quantum computing. At a high level, they offer fundamentally different capabilities compared to classical, digital computing machines. These capabilities include simulating natural phenomena using physics as a direct model. The urgent challenges in harnessing these promising models of computation are in connecting them to conventional architectures, and in helping programmers correctly use them.First, I will talk about how analog accelerators can play a role in modern architectures to aid in modeling stochastic and nonlinear phenomena. The key feature of the analog model of computation is that variables evolve continuously, thereby avoiding many problems associated with the step-by-step updating of variables in all digital machines. Using prototype analog accelerators that I helped build at Columbia University, I demonstrate using analog approximate solutions as good initial seeds for GPUs solving scientific computation problems.Second, I will talk about helping programmers debug and simulate quantum computer algorithms. As more capable quantum computer prototypes become available, students and researchers need guidelines and tools for debugging and testing quantum programs. My work discusses what quantum bugs are, how statistical tests may find them, and how programmers can write assertions to more effectively bring up quantum programs. Furthermore, I show how tools originally meant for classical inference may enable new queries in aid of debugging quantum programs.

: 

Contact  Faculty Host: Santosh Nagarakatte