NERSC is inviting proposals for projects that will leverage NERSC’s Perlmutter supercomputer to push the state of the art in Generative AI (GenAI) and deep learning for science and produce novel science outcomes. We are specifically seeking teams with expertise using deep learning for science, a deep understanding of the scientific domain, and demonstrated proofs-of-concept. NERSC staff effort will be available for consulting on running effectively at NERSC, but not generally for model development.
Initially, O(10,000) Perlmutter GPU node hours will be available for each project (each Perlmutter GPU node contains four A100 GPUs) with associated storage quotas on NERSC’s filesystems. Additional resources may be available for projects that can demonstrate their ability to effectively utilize them.
Apply:
Fill-out the application form for consideration. Proposals will be reviewed on a rolling basis. Submissions made by April 1, 2024 will be given full consideration.
Contact Shashank Subramanian or Wahid Bhimji for further information.
Guidelines:
Interested teams are invited to submit a proposal using the application form that includes:
- A well-defined project scope that includes relevant scientific background, objectives, methodology and impact to the scientific community
- Computational resource requirements and utilization plan
- Team expertise, experience as well as clear and specific deliverables along with an expected timeline
Criteria:
Projects will be evaluated based on the following criteria:
- Scientific significance and innovation
- Degree of relevance to the missions of NERSC, Berkeley Lab and the DOE Office of Science
- Technical feasibility of the proposed project
- Potential and readiness for effectively leveraging NERSC’s large computational resources at scale
- Clarity in scope, timeline, objectives and track record of the research team
Awards are for the NERSC 2024 Allocation Year, which runs through January 15, 2025. Projects will be expected to use their allocation within a six-month period, or be subject to a reduction in allocation. Awardees will be required to report progress and summarize achievements to NERSC.
