Dear users,

We are pleased to announce the addition of three new compute nodes to the boost_usr_prod partition: aldo, giovanni, and giacomo. Each node is equipped with 2 NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, featuring 96 GB of VRAM per GPU.

These nodes will be primarily dedicated to running LLM inference services for the entire University. To target these nodes specifically, use the new SLURM feature tag:

–constraint=”gpu_RTXPro6000B_96G”

As usual, if your job only requires a minimum amount of VRAM, we recommend including all features that satisfy your requirement to maximise scheduling opportunities.

For more details on node features and partition limits, refer to the documentation.

Best regards,
Lorenzo.