GPU “Just in Time” Compiler (aka “Shader Compiler”)

The Computer Science Department & the Center for Computer Systems Research will sponsor a seminar Friday, November 4, 2011, in Rekhi Hall, Room 214 featuring  Dr. Gang Chen.  Dr. Chen from AMD Boston Design Center,  will talk about the just-in-time compiler shipped with every AMD GPU and APU product, what it is and why it is an essential component of the GPU. The talk will include a brief overview of GPU architecture, focusing on the differences from CPU, a brief overview of the JIT’s internal structure, and interesting compiler issues that may not be seen with a typical CPU compiler along with some tips about performance tuning, and some of our ongoing and future works.

Gang Chen is currently the lead engineer for AMD’s GPU just-in-time compiler, and responsible for developing a new compiler for AMD’s next-generation GPU. He has been working in this field for AMD/ATI since 2002. Before that, he worked on a DSP C compiler at StarCore Technology Center for Agere from 2001 to 2002. He received his Ph.D in Computer Science in 2001 from Harvard University, his BE and ME in 1991 and 1993 from Tsinghua University, Beijing.