Invited Panel

Will AI be all we need to program FPGAs by 2030?

Moderator

Gabriel Weisz (AMD)

Panelists

  • Andrew Putnam (Microsoft)
  • Jose Renau (UC Santa Cruz)
  • Rama Venkata (Altera)
  • Zhiru Zhang (Cornell)

Abstract

LLMs are useful for helping application developers write code for all sorts of problem domains and have even been used for chip design. How will this change application development for FPGAs? Will we be able to ask ChatRTL to accelerate itself on a Stratix 9000 or Versal Supreme FPGA? Or will AI be limited to creating unit tests and small modules?