With the final delivery of all remaining hardware components on May 12th, implementation of the new ManeFrame II (M2) High-Performance Compute Cluster enters the homestretch.
This new cluster of computing power increases our research computing capacity from 110 teraflops to 610. This improvement sets the stage for the implementation of additional features in a future second phase, including graphical desktop nodes, a simpler job management interface for researchers, and archival storage. Under the hood, ManeFrame II packs a punch with Intel Phi accelerator processors, NVIDIA P100 GPUs, and additional high-memory compute nodes. It’s all connected with an InfiniBand low-latency 100 gigabit network.
The hardware setup in the University Data Center is complete, and exhaustive benchmarking and acceptance testing is currently taking place. Once complete, the Center for Scientific Computation and OIT hope to bring this revolutionary new computing platform to our research community by the end of the Summer.
ManeFrame II Upcoming Milestones:
| May 12 – May 31 | SMU benchmarking and acceptance testing process |
| June 1 – June 9 | Reset the cluster for SMU production work |
| Week of June 12 | Begin transition for early adopters |
| Summer | Transition all users & data from M1 to M2 cluster |
| Fall | Decommission old cluster pending Faculty availability |
For further rollout details and technical specifications for the ManeFrame, we have a full M2 report available on Box at smu.box.com/v/m2-techspecs.
