The Discovery infrastructure has undergone a substantial expansion, now comprising over 6,000 computational cores on CentOS linux. This system is further bolstered by an array of GPU nodes that incorporate 10 A100 GPUs, 72 A5500 GPUs, and 12 V100 GPUs. These resources interface with a 2.6 PB Isilon storage system, denominated as DartFS. This computational platform is ready to support the research needs of the Dartmouth academic community, including those pursued within the Geisel School of Medicine and other the professional academic schools.
Discovery offers extensive support for compilers, encompassing those from Intel, as well as the C and FORTRAN languages. We warmly invite requests for the installation of additional software and ask that these be directed to research.computing@dartmouth.edu. For the efficient management and allocation of computational resources, job submissions on Discovery are channeled through a job scheduler, Slurm. This method ensures an equitable distribution of resources and optimal CPU usage. For guidance on scheduling jobs, please refer to the Scheduling Jobs to Run
Access to the Discovery cluster is limited to those on the on-campus network or those connected to the Dartmouth VPN. Users can log in using clients that support SSH, SFTP, or similar protocols. To sign up for a Discovery account, you can fill out the account request form: Click here.
We actively encourage research groups with demanding computational needs to explore the possibility of acquiring additional nodes. For insights on this, the Investing in Discovery page offers more details. For a wealth of articles on Dartmouth’s ITC Knowledge Base, see this for Details or simply search ‘Discovery’ at services.dartmouth.edu.
Our immense gratitude goes to the Office of the Vice Provost of Research and Information Technology Services for their generous institutional backing.