We are expanding the Peregrine system to include an additional 1152 nodes. Each node will have two 12 core Intel Xeon (Haswell) processor chips. This will add an additional Petaflop of peak performance, brining the overall peak performance of Peregrine to 2.24 Petaflops. Each node will have 24 cores and 64 GB of DDR4 memory. These nodes will be organized into two tightly connected groups of 576 nodes, which will increase the maximum number of nodes per job from 288 to 576. We plan to have these nodes available to users in early November.
last modified Sep 29, 2015 11:35 AM