Hyperscale clouds, consisting of millions of standard 2-socket CPU servers, are the backbone of modern computing. Sustainably growing the cloud’s computing capability requires major advancements in efficiency, with specialized hardware accelerators promising massive improvements in performance and power savings. While some accelerators for storage and networking have made their way into the cloud, the landscape of general-purpose computing accelerators remains limited. With the advent of AI, GPUs have emerged as the dominant accelerator. Field Programmable Gate Arrays (FPGAs) have long been utilized for both compute and network/storage acceleration, but their role in the future of hyperscale clouds is now in question. Will GPUs continue to be the sole relevant accelerator, or is there still a place for FPGAs or other chip architectures in the evolving hyperscale cloud infrastructure?
This presentation will critically examine hardware accelerators in the cloud, focusing specifically on the FPGA industry’s influence and its triumphs and setbacks. We will discuss obstacles to cloud adoption of accelerators, why software is more important than hardware in driving accelerator adoption, and identify pivotal research areas that could propel FPGAs toward groundbreaking innovation in specialized hardware and accelerated computing alongside GPUs.
Andrew Putnam is a leading figure in engineering hardware accelerators for the hyperscale cloud, serving as a Partner General Manager of Cloud & AI Hardware Engineering within Microsoft’s Azure Hardware Systems & Infrastructure (AHSI) group. He co-founded the Microsoft Catapult project, where his pioneering work integrated Field Programmable Gate Arrays (FPGAs) into production hyperscale data centers, effectively doubling server capacity for web search. His efforts have been instrumental in developing Azure Accelerated Networking, Project BrainWave, and Azure Boost, contributing to some of the fastest network, storage, and AI solutions in the cloud, and showcasing how PhD research can translate into massive industry impact at scale.
Dr. Putnam's academic journey is distinguished by a unique triple major in Physics, Computer Science, and Electrical Engineering from the University of San Diego, followed by a Master’s and Ph.D. in Computer Science and Engineering from the University of Washington. He began his career at Microsoft in 2009 as a researcher, quickly gaining recognition for his innovative contributions in FPGAs and computer architecture. In 2016, he transitioned his research into practical applications within Azure Networking, leading to his current leadership role in Azure Hardware since 2020.
