Inside look at Oak Ridge National Lab’s facilities for Summit -> Frontier; The Quantum Internet; ‘ADIOS’ to Data Management; AI for Neuroimaging; Cloud Price/Perf for Genomics; + More..
On today’s program we get an inside look at the Oak Ridge National Lab with facilities director, Jim Rogers. We take a look at the computing facility but also step back to see the power and cooling infrastructure. During the interview the focus is on the lifespan of datacenters at ORNL, how those are extended, and how the teams who build leading supercomputers think about efficiency and longevity. For a more self-guided experience, here’s a virtual tour.
We also talk about the future of the quantum internet with ESnet director, Inder Monga. We look at the DoE’s investment in the concept and the technical challenges that await on the networking and applications front over the next several years. In addition, we also discuss the role such a network might play for future scientific and technical computing.
In addition, we look at solving data management challenges in HPC and large-scale computing with Oak Ridge National Lab’s Scott Klasky with the ADIOS-2 framework. We walk through the evolution of the platform and look toward exascale challenges where such tooling might be more useful than ever.
We get some outstanding price/performance analysis of cloud for genomics in this episode, complete with some charts of interest. The instance type and cost information we discuss with our University of Sheffield and Microsoft Research guest will be useful for any data-intensive workload and is definitely worth a closer look.
Also on the program we talk with researcher Sergey Plis about the role AI will play in future neuroimaging and neuroscience efforts via the NoBrainer project. More specifically, we talk about a platform his team at Georgia State University has developed that trains a complex neurological AI model with the goal of handing this over for more specialized, tuned training at other research centers.
Thanks as always for tuning in.