More GPUs plus 1.5 TB RAM Nodes
The Research Lifecycle Programme Delivers Again!
Two months ago our HPC Pool — 4096 CPU cores of Infiniband-connected compute resource all dedicated to true HPC work — went into production. Many people are already running jobs on The Pool and this platform was delivered via RLP Change Project Z: HPC Investment.
More new compute resources will be available within days! Through RLP Change Project M&K (Incremental Increase in Compute Resource and Orchestration), 20 state-of-the-art Nvidia V100 GPUs will be made available to CSF users to share, from a total of 44 (with more coming soon) — the extra 24 were contributed by research groups who will get priority access to those. There will also be three very high memory (1.5 TB RAM) nodes and some additional standard compute nodes and some Cloud-based resources.
Thanks very much to those who attended the M&K workshops back in the autumn. Output from these determined what we procured. (Output from the recent M&K workshops is feeding into M&K Phase 2, for 2019/20.)