[PAST EVENT] AI, app-defined OS, and the ossified Linux kernel
Speaker: Felix Xiaozhu Lin (https://engineering.purdue.edu/~xzl/), Purdue University
Title: AI, app-defined OS, and the ossified Linux kernel
In response to AI and big data, systems software evolves in two directions: many classic OS functions now incarnate as app-specific runtimes; the kernel beneath them is deprived of personalities and ossified. We address challenges along both directions. I will describe our recent work on supporting large stream and video analytics as well as on building unconventional kernels. I will discuss the unified themes underlying the two seemingly distinct inquiries and the road ahead.
Felix Xiaozhu Lin is an assistant professor of Purdue ECE. He cares about systems software for computers of which power consumption in the (10mW, 500W) range. He takes diverse approaches, including novel OS structures, kernel subsystem design, whole-OS characterization, binary translation, and user-level runtimes. See http://xsel.rocks for his projects, one of which (K2) won the best paper award of ASPLOS’14.
He received Ph.D. in CS from Rice University, BS/MS from Tsinghua University, and a Google Faculty Award.