Skip to content

FCC Analysis Facility at BNL

BNL operates an analysis facility available to all FCC physicists and computer scientists. The facility currently provides:

  • ~100 CPU nodes for general computing
  • 10 NVidia L40S GPUs for machine learning and vectorized computing
  • Disk storage under /gpfs/mnt/gpfs01/usfcc, plus ~100 TB of common space for sharing datasets

The system architecture and management model will be based on the ATLAS Tier-3 computing environments. It is being organized and maintained to support the computing needs of FCC users, including capabilities such as:

  • Logging into interactive sessions
  • Running batch and interactive jobs
  • Accessing and storing both shared and private data

As FCC workflows evolve, the facility will be continuously updated to support new requirements.

Tools and Software Support

The Analysis Facility (AF) supports a wide variety of tools tailored for data analysis. These include:

  • CERN software via CVMFS
  • Grid middleware
  • Rucio clients
  • Machine learning packages
  • MPI (Message Passing Interface)
  • JupyterLab with PyROOT
  • XCache with automatic data discovery
  • GPU acceleration, and more

Getting Started

New to the facility? Start with the Onboarding Checklist. It walks you through account application, MFA, SSH keys, first login, software setup and your first batch job in order.

Getting Help

What you need Where to go
Account creation and access problems rt-racf-useraccounts@bnl.gov
Questions, discussion, announcements AF Discourse Forum — see the "BNL FCC analysis facility" section
Facility requests and feedback sabidi@bnl.gov

Documentation Index