Five EECS faculty receive Ted Kennedy Family Faculty Team Excellence Award

These faculty each contributed toward a successful center that developed specialized hardware “building blocks” for a range of applications.
Kennedy Team award winners
Austin, Bertacco, Das, Kasikci, and Lu

Five EECS Faculty affiliated with the Applications Driving Architectures (ADA) center have been awarded the Ted Kennedy Family Faculty Team Excellence Award for 2022-23. The recipients include Profs. Todd Austin, Valeria Bertacco (center director), Reetuparna Das, Baris Kasikci, and Wei Lu.

The ADA center is a $28M center based at the University of Michigan that engages ten universities, 20 faculty members, and 130 graduate students and that has been active from 2018-2023. ADA’s goal has been “to streamline and democratize the design and manufacturing of next-generation computing systems” through the research collaborations of the participating faculty and industry sponsors.

At the launch of the center, very few specialized architectures were available, primarily for neural networks. Throughout the past years, the research team developed a rich ecosystem of hardware accelerators spanning many critical domains, from media processing, to data privacy and graph-based computation, to explainable AI algorithms. A key aspect of the ADA center effort was to adopt emerging silicon technologies and deploy them in novel system solutions at scale. Several of the PIs developed silicon prototypes to deliver proof of concepts of their ideas, including new types of memory devices.

On the hardware design front, researchers devised new hardware design flows that allow the compilation of an application specified with a high-level language directly into an accelerator rich heterogeneous hardware system. This latter part of the research is only possible because of the frequent discussions involving computer architects, compiler researchers and programming language experts – each of these groups traditionally work in distinct communities, interacting rarely across disciplines. The work of the research center not only brought them together, but promoted cross-disciplinary brainstorming and solution development.

In addition, the center created an ecosystem of collaboration and interactions between researchers and industry sponsors, including weekly “liaison meetings,” where students would discuss their ongoing research, and semi-annual get-togethers and brainstorming. The graduate students participating in the center got to know industry technologists, faculty and other doctoral students from participating institutions, thus enjoying a high level of interactions with their future community of scholars. Several of these students would, in turn, spend internships and/or be hired after graduation by an industry sponsor, thus facilitating the transfer of ADA technology into industry. Several other forms of technology transfers have been enabled through the launch of multiple startups spun out from ADA research, including Agita Labs (Austin/Bertacco), MemryX (Lu), and Sequal (Das).

Thomas Wenisch, Director of Engineering at Google and a former faculty member in EECS, made the following remark about ADA: “Organizing a research enterprise of this scope is an enormous endeavor. The Michigan leadership team, headed by Valeria, repeatedly pulled off wonders in organizing research symposia, webinars and industry liaison, technical exchanges with DARPA and sponsors, while meeting a complex set of reporting and research dissemination requirements.”