Compute Infrastructure We Use¶

Overview¶

Overall Infrastructure Chart with system specifications and notes.

The Ohio Supercomputing Center (OSC): Large compute resource accessible through ssh or OnDemand (web) platform.
- Globus Endpoint — this is the preferred transfer method for collaborations, especially for large data.
- OnDemand page: info, login
- myOSC client portal: info, login
Imageomics dedicated GPU server: Internal server, hosts our CVAT instance.
- Usage and access guide (internal)
- CVAT user guide
NSF ACCESS Accelerate Allocation: NCSA Delta GPU credits
Amazon Web Services (AWS): Basic to extremely powerful, abundant (though finite) resources, high cost.
- Used sparingly for urgent deadlines when other compute is not available (generally hasn't been available at those times either, though) or to host projects that cannot be hosted effectively through a Hugging Face Space.
- AWS usage guidelines (internal)