Berkeley Research Computing

Berkeley Research Computing (BRC) logo

The Berkeley Research Computing (BRC) program supports research with a coordinated set of services across a range of computation and data analysis needs. BRC includes consulting that ensures faculty are matched to the appropriate resources, and that the computation and data resources meet their needs. The BRC program is sponsored by the CIO, the Vice Chancellor for Research, and the Chancellor.

Overview | Faculty Perspective | Service Roll-out

Please tell us how BRC impacts your research (Google Form)

BRC Overview

(one-page overview, PDF)

Consulting and Community

  • First point of contact for faculty and other campus researchers, to ensure match to appropriate resources
  • Specialist consulting in each service area
  • Support migration between computational modes for efficient scaling and reproducible research
  • Domain consultants for specific skills, as part-time staff or Fellows
  • Foster cross-domain community and knowledge-sharing
  • Partnering with network of partners (D-Lab, SCF, BIDS, et al.) to coordinate consulting services across campus

Savio Condo/Institutional Cluster

  • High performance computing with high-speed, low-latency interconnect and high-speed parallel filesystem
  • Standard, High-Memory, GPU, and High-Throughput-Computing (HTC) pools available
  • Infrastructure and admin costs subsidized for Condo contributions, unused capacity available to campus
  • Condo users can “burst” beyond their hardware
  • Faculty computing allowance provides free access for PI’s
  • Supports students learning computational methods
  • Managed in partnership with LBNL HPC Services group

Cloud Computing Support

  • Consulting services and documentation to solve common problems of access and administration of public cloud resources, and integration with teaching
  • Support for commercial services (AWS, Azure, etc.) as well as national centers (XSEDE, NERSCSDSC, etc.)
  • Integration with data and network infrastructure
  • Investigating flexible cloud-like computing environments within Savio
  • Working with EECS, AMPLab, D-Lab to develop materials and services

Analytics Environments On-Demand (in development)

  • Virtual workstations: self-serve access to compute, storage, and commonly used software
  • Supports research workflows in the social sciences and humanities 
  • Domain-specific VM images can be prepared, managed
  • Anywhere, anytime access to lab workstation model
  • Working with Library, D-Lab, ETS, and IST to provide service infrastructure and consulting

 

Computation and Data

BRC consultants work closely with the Research Data Management (RDM) program to support data and computational needs across research workflows.

 

BRC Program Launch

On 22 May 2014, the Berkeley Research Computing Program was formally launched (iNews article). Program Manager Patrick Schmitz gave the following overview of the program at the launch event held at Sutardja Dai Hall:

Faculty Perspective

UC Berkeley History Professor Cathryn Carson, Interim Director of the D-Lab and Associate Dean of Social Sciences, speaks at the BRC Launch of 22 May 2014 on the diversity of research among 450 faculty in Berkeley's social science disciplines, as well as ethnographic similarities as new computational methodologies are taken up across these disciplines. Prof. Carson identifies types of computational resources required by social scientists, and describes several research projects that make use of those resources. She describes Berkeley Research Computing Program staff engagement with faculty and graduate students in order to frame the program in the context of needs articulated by the researchers themselves.

 

BRC Services roll out

BRC Services are rolling out in sequence, with the Savio (HPC) service first in production, followed by consulting services and Cloud Computing support. We are investigating support for additional modes of computation and best practices for mobility of computation across diverse computational resources.

 

Partnership: