SeaWulf is Stony Brook's general-purpose cluster: hundreds of CPU nodes, high-bandwidth and high-memory nodes, and A100 GPUs, open to the campus community and Brookhaven Lab staff.

400+compute nodes
23,000CPU cores
1.86PFLOP/s peak performance
3 TBRAM on the largest nodes

Get connected

In your browser

Open OnDemand

Jupyter, RStudio, MATLAB, a full desktop and 14 more apps, with nothing to install. Sign in with your NetID and Duo.

Launch Open OnDemand
From a terminal

SSH

The milan login nodes reach most queues. Use login.seawulf.stonybrook.edu for the 28-core and older GPU nodes, and xeonmax for HBM.

ssh <netid>@milan.seawulf.stonybrook.edu

Windows setup, Duo tips and every login address: Connecting to the cluster.

Specifications

Node typeDetails
40-core Intel Skylake192 GB memory per node, AVX-512
96-core AMD EPYC Milan256 GB memory per node, AVX2
High-bandwidth memoryIntel Sapphire Rapids, 256 GB DDR5 plus 128 GB HBM (one 1 TB node)
A100 GPU nodes4 NVIDIA A100 80 GB GPUs per node, 64-core Ice Lake, 256 GB memory
High-memory nodesUp to 3 TB RAM
28-core Intel Haswell128 GB memory per node (legacy)
Older GPU nodesK80, P100 and V100 (legacy)

Which queue?

Your jobTry this queue
A quick test, under 1 hourdebug-40core
A few cores, not a whole nodeshort-40core-shared or short-96core-shared
Whole nodes, up to 4 hoursshort-40core or short-96core
Up to 12 hoursmedium-40core or medium-96core
Up to 2 dayslong-40core or long-96core
Up to 7 days, 1 to 2 nodesextended-40core or extended-96core
Needs a GPUa100, or NVwulf for large AI work

Every queue, its limits and where to submit it from: Queue selection guide. Or let the job script builder write the script.

Storage

SpaceSizeBacked upGood to know
Home /gpfs/home/<netid>20 GBYesScripts, settings and small files
Scratch /gpfs/scratch/<netid>20 TBNoFiles older than 45 days are deleted
Project /gpfs/projects/<group>Up to 10 TBNoShared by your group, on request

Details and how to ask for project space: Storage layout & policies.

Popular apps in your browser

Best suited for

  • Physics and climate science
  • Bioinformatics and materials science
  • Machine learning and AI
  • Engineering and data science

Getting access

  1. Join a projectIf your PI already has a SeaWulf project, open a ticket with the help topic "Request an account for SeaWulf" and include your NetID, your PI's name and the project number.
  2. Or start one (faculty)Faculty request a project number with a one-page PDF describing the work, computing needs, software, storage and group members.
  3. Set up Duo, then connectEvery login needs a Duo approval. Then sign in with Open OnDemand or SSH and run your first job.

Related guides