Alternative Computing Technologies Lab

Welcome to the Alternative Computing Technologies (ACT) Lab. Our lab is dedicated to advancing the field of computer science and engineering by developing innovative technologies and cross-stack solutions that enhance the performance and energy efficiency of computer systems, particularly for emerging applications such as transformers. At ACT Lab, we take pride in our commitment to open-source initiatives. We have pioneered several open-source projects, including our compilation framework - Polymath, full-stack DNN compiler - Codelet, a unified template-based framework for accelerating statistical machine learning - Tabla, and DNN acceleration systems generator - GeneSys. These projects reflect our dedication to collaborative research and our belief in the power of open-source contributions to drive innovation. Our current focus extends beyond pioneering work in enabling accelerators within the system stack to facilitate cross-domain multi-acceleration for next-generation datacenters. Our research group actively drives innovation into cutting-edge domains that defy convention, including the development of brain implantable devices and the advancement of privacy-preserving machine learning techniques. Our dedication to pushing technological boundaries and exploring novel applications reflects our commitment to advancing the frontiers of research.