A central theme at Tuesday’s AI Design Forum, part of SEMICON West held in San Francisco this week, was how the semiconductor industry is reacting to the slowing of Moore’s Law, and how design innovations augmenting performance are enabling powerful, data-intensive applications in the areas of AI, IoT, 5G, big data and automation. Xilinx CEO Victor Peng was a featured speaker, offering his perspective in a keynote as well as a panel discussion alongside CEOs from across the industry including Lisa Su of AMD, Aart de Geus of Synopsys, Gary Dickerson of Applied Materials and Sanjay Mehrotra of Micron Technologies. Together the CEOs shared their respective views on how computing is rapidly becoming less centralized and more embedded, interconnected and flexible to serve increasingly complex needs across different verticals.
CEO Technology RoundtableFrom the keynotes, the overall message fell along the same lines: compute is expanding inside and out of the data center, across networking technologies and to the edge at connected devices. Innovation today means placing agile, high-performance computational capabilities everywhere. At Xilinx, the future is already here.
On stage, Victor highlighted the defining innovations Xilinx is implementing today that are driving forward unheard levels of performance and will deliver greater value for industries from agriculture to automotive, even as Moore’s Law loses steam. He spoke to the importance of new adaptable computing approaches specifically around the ACAP category of devices Xilinx created that far exceeds GPU capabilities. Notably, he discussed recent projections for our Versal ACAP technology in its processing power for neural networks. Versal ACAP processed an extraordinary 9,000 to just under 12,000 images per second with sub-2ms latency. For context, the latest GPU on the market achieved just 2,600 images per second with sub-2ms latency. That’s 3x to 4.5x the performance.
As industries look to AI to drive innovations, the AI Design Forum provided a strong platform to reach important developers building next-generation solutions. AI holds so much potential, yet it is still in its infancy and we are committed to accelerating this vital evolution in computing.