Abstract: Hardware accelerators, especially GPUs, built to sustain the progress in AI are often underutilized, barely reaching 50% utilization, while running deep learning workloads. This waste of hardware resources is exacerbated by the expensive price of these accelerators, while contributing to the carbon footprint of AI systems. To close this utilization gap and reduce the hardware resource needs of deep learning, the DEEP project aims at building a collocation-aware resource manager for deep learning workloads running on GPU clusters. This talk will give an overview of the steps planned for achieving this goal.
Short Bio: Pınar Tözün is an Associate Professor and the Head of Data, Systems, and Robotics Section at IT University of Copenhagen (ITU). Before ITU, she was a research staff member at IBM Almaden Research Center. Prior to joining IBM, she received her PhD from EPFL. Her thesis received ACM SIGMOD Jim Gray Doctoral Dissertation Award Honorable Mention in 2016. Her research focuses on resource-aware machine learning, performance characterization of data-intensive systems, and scalability and efficiency of data-intensive systems on modern hardware.
Personal website: https://www.pinartozun.com
Host: Prof. Pamela Delgado
