Invited Panel
Do FPGAs Still Have a Place in the Era of AI?
Moderator
Andrew Putnam (Microsoft)
Panelists
- Seong Kim (AMD/Xilinx)
- Raymond Nijssen (Achronix)
- Greg Stitt (Quantlab/University of Florida)
- Bita Darvish Rouhani (NVIDIA)
- Eriko Nurvitadhi (MangoBoost)
- Ilya Ganusov (Intel/Altera)
Abstract
FPGAs have played a significant role in the computing industry, but they thrived by being different — addressing problems domains that other technologies couldn’t, or at least being the earliest to the party. During eras where the computing industry was focused on mainframes, PCs, and mobile, FPGAs found success in domains like ASIC prototyping, telecom, networking, and low-volume specialized compute.
The computing industry has shifted focus to Cloud and AI, but instead of being different, the FPGA industry and FPGA conference publications are often concentrating on those same areas. When the biggest players in computing are focused on the same things as FPGA researchers and developers, will the FPGA industry still continue to provide outsized impact? Or will larger and better-funded competition simply outpace the traditional advantages of FPGAs?
This panel discussion will explore the role of FPGAs in the computing industry in the new era of the Cloud, Large Language Models and AI. The panelists will explore:
- How they found their niche in prior eras
- Whether the industry is having the expected impact in Cloud and AI
- Where they would look if they were starting anew
- Whether the FPGA industry and researchers are focused on the right problems
As usual the panel is a space to explore the design space in a casual yet thoughtful manner, with tough questions, honest answers, and humor particularly appreciated.