Next generation Xilinx architecture and software

Wednesday September 26, 2018
Location: Hamerschlag Hall D210
Time: 4:30PM-5:30PM

The talk will start with a brief overview of Xilinx's current architecture and software stack for Machine Learning, Embedded Vision, Automotive etc. The talk will then motivate the next-generation Adaptive Compute Acceleration Platform (ACAP), based on Xilinx's broad focus areas spanning embedded to data center applications. Further, some details of this new architecture, which includes a large array of programmable compute cores, will be presented. The talk will conclude with a description of the programming environment for this heterogenous compute architecture, and the challenges and opportunities that lie ahead.

Mukund Sivaraman got his PhD from Carnegie Mellon University in the area of timing analysis in 1997. He subsequently did research in VLIW processor architectures and compilers at Hewlett-Packard Labs, then co-founded Synfora, a high-level synthesis startup, in 2003 that was acquired by Synopsys in 2010. Mukund is currently at Xilinx leading a group developing programming and compiler tools for the next-generation Xilinx architecture.