← All Updates
Scaling Run Completed: Hyperparameter Optimization on Up to 192 GPUs
Professor Kay Giesecke’s team, in Management Science and Engineering, completed a scaling run on Marlowe across up to 24 nodes (192 H100 GPUs) for hyperparameter optimization ahead of a larger campaign. The run began Friday, July 10 at 3:00 PM PT and ended Monday, July 13 at 7:00 AM PT — approximately 64 hours.
The run was announced to Marlowe users on July 8. For its duration the batch partition was drained but continued to accept job submissions to the queue, and jobs running in batch at 3:00 PM Friday were requeued and eligible to resume from 7:00 AM Monday. The preempt partition ran as usual.