This summer we continued to grow and evolve Research Computing on campus and across the system. We would like to welcome Predrag Lazic to the group. Predrag comes from a career in academia where he worked in the fields of computational materials science and physics and has expertise in large-scale parallel computations. We would also like tothank Jacob Gotberg for his work as he has left the University.
We have been working behind the scenes to grow and improve the infrastructure that supports the clusters. The Clark cluster was upgraded this summer with a new login node, head node, and four compute nodes (96 cores) on upgraded hardware. This was done in part to make room for the new hpc6 rack and to test our next generation cluster management environment.
We designed and ordered the next generation of compute capacity (hpc6), the nodes are expected to arrive mid October. The nodes contain the latest Intel processor (6th generation Skylake) and Infiniband network (Melenox HDR200) technologies. This system allows us to put 64 nodes with 3072 cores in a single rack. We will be adding 32 nodes (1536 cores) in Phase 1 with room for 32 more nodes (1536 more cores) to be added in Phase 2.
The new rack is being added in anticipation of older community nodes (hpc3) reaching end-of-life in January 2020. These nodes (1152 cores) will be removed and/or re-purposed sometime after that. In October 2021 the hpc4/hpcr4c (2044 cores) will reach end-of-life and will not be replaced. A large number of these nodes were purchased as part of individual investors (thank you!) and large grants (an engineering lead NSF MRI and a Mizzou Advantage grant for Bioinformatics) and are not a part of community compute capacity (the general account). This means that the overall capacity of cluster will be significantly reduced at that time unless there is more investment by the community. This will put a lot more pressure on jobs from users that have not invested (running under the default general account and General partition) and will probably make it a lot more difficult obtain a lot of compute time without investment.
Now is the time to start writing your computational and storage needs into your grants. We have template information to easily include in your grant proposals and a number of grant-friendly options for investment.
We are also proud to announce that we are a part of a 1.4 million dollar NSF CC* CyberTeam grant (Award # 1925681) to help grow and evolve Cyberinfrastructure in the region. This grant will give us opportunity interact and learn from other research computing centers in the region.
You can find more information at
— Oct. 1, 2019