USCMS Researcher: Patrick McCormack
Postdoc dates: Jun 2021 - Jun 2023
Home Institution: Massachusetts Institute of Technology
Project: Accelerating offline computing with the Fast Machine Learning Lab
With the upgrade to the High Luminosity LHC, the demands for offline computing resources in CMS will grow immensely. Both additional computing resources and new ideas for how to process the data are needed to ensure that CMS can continue to produce high quality scientific results. The recently developed SONIC framework simplifies the usage of accelerators such as GPUs and FPGAs in both offline production and triggering within the CMS computing workflow. Our team aims to turn heterogeneous computing into a major research effort within CMS.More information: My project proposal
Mentors:
-
Philip Harris (MIT)
- 2 Nov 2021 - "MiniAOD Production with Inference as a Service for ML Algorithms", Patrick McCormack, CMS S&C R&D Postdocs Meeting
- 13 Sep 2022 - "Heterogeneous architectures via inference as a service in CMSSW", Patrick McCormack, CMS S&C R&D Postdocs Meeting
- 6 Feb 2023 - "Integrating SONIC for accelerated inference as a service in CMSSW", Patrick McCormack, CMS S&C R&D Postdocs Meeting
- 11 May 2023 - "Portable Acceleration of CMS Mini-AOD Production with Coprocessors as a Service", Patrick McCormack, CHEP 2023
Current Status
Contact me: