Further Bianca hardware upgrade closed

Bianca has got an upgrade with more GPUs and memory.

17 nodes with Nvidia T4

Seventeen nodes in Bianca have been upgraded with the remaining 17 Nvidia T4 GPUs. To use the nodes start jobs with --gpus=t4:1.

The ten nodes with dual Nvidia A100 GPUs remain available. To use them start jobs with --gpus=a100:2. Since A100 is the default this also works: --gpus=2

The old method of allocating GPUs on Bianca (--gres=gpu:2 -C gpu) also works to get an A100.

The Nvidia T4 GPUs have been repurposed from the old decommissioned cluster Snowy.

4 nodes with 1 TB RAM

Four nodes in Bianca have been upgraded to 1 TB RAM each. To use these nodes start jobs with the option -C mem1TB. This feature is new for Bianca, previously 512 GB RAM was the largest available memory size.

The RAM has been repurposed from the old decommissioned cluster Rackham.