HCC just installed 56 new nodes featuring Intel Omni-Path (OPA) interconnect, a 100Gb/s high-speed link technology, to the current Crane cluster.
Compared to the original Crane compute nodes, the newly added nodes have much faster CPUs, 225% more cores per node, 300% larger memory (256GB or 512GB per node), and 150% faster networking using 100Gb/s Intel Omni-Path interconnect. These new nodes not only deliver better performance for computational research but also enable large memory calculations, which was only available previously on our older and slower cluster, Tusker.
The new nodes are available now for general HCC use and for dedicated group ownership. Though they use an architecture different from Infiniband, users’ jobs on submitted to the Crane batch partition can be scheduled to these new nodes, and automatically benefit from the 100Gb/s.
Moreover, multiple projects are on-going which will further strengthen our computing power and expand the research resources we offer. New compute nodes, GPU nodes, storage, networking, are in either ordering, purchasing, or planning phases, making this an ideal time for buy-ins.
If you have any questions about our new nodes or are interested in purchasing dedicated resources, please contact hcc-support@unl.edu or see current pricing at http://hcc.unl.edu/priority-access-pricing