SC ’15, the big, annual High Performance Computing (‘HPC’) conference, officially kicks off in Austin this evening with a gala opening of the conference exhibits. As we did last year, VMware has a demo booth within the EMC booth (#499) on the show floor. When I mentioned this to a customer recently, he cleverly asked whether the VMware booth would be embedded within the EMC booth embedded within the Dell booth at SC’16 in Salt Lake City, like Matryoshka nesting dolls. I don’t know about that, but I do know we have some interesting things to present at this year’s show.
We will be highlighting our High Performance Analytics solution in the booth this year. This solution is an abstracted interface that allows researchers to focus more on their science. This means that much less attention needs to be paid to the underlying IT infrastructure used to run their analyses. We will also be presenting the latest performance information for HPC applications running on our virtual platform, including the latest MPI latency numbers.
We will give two talks in the booth theatre this week. Ranjit Sawant will talk about HPA in a talk titled, “Research Collaboration and Scientific Agility using VMware High Performance Analytics”. In addition, I will talk more generally about our overall HPC approach, including information about performance, in a talk titled “VMware Private Cloud for High Performance Computing”.
We’ve also just completed a case study detailing work done by Edmond DeMattia at the Johns Hopkins University Applied Physics Laboratory to virtualize their Monte Carlo clusters, which yielded significant increases in resource utilization and better application performance than native. It’s a great example of how virtualization can bring value to HPC environments, and we will have advance copies of the report available in the booth.
Please stop by to learn more about what VMware is doing in HPC, to talk in detail about Infiniband/RDMA performance results, or to learn why customers are interested in building private cloud environments for running their HPC workloads.