HPC Server SP1: Burst to Azure with little or no programming
We all know that Scientists and Engineers run Computationally intensive jobs. Many of us often wish we had an extra computer or a dozen to run the applications that seem to take forever on our desktop. We wait for the code to compile, the simulations, experiments to run, and quite often we let our jobs run overnight and come back to check the results in the morning. Computers have gotten faster over the years, but all it takes is for the user to add a few digits to the problem size to eat up computing power we have. As the processor clock speed hit their 3 ghz limit a few years ago, computers are just not getting faster, fast enough.
These technical computing applications or *any* compute intensive executable fit naturally with the Windows Azure business model: on-demand CPU intensive workload by the hour. I would like to introduce you to the HPC Server Sp1 product with Azure burst capability. Once setup, you would be able to run your typical CPU intensive jobs on Azure without programming when local CPU are already occupied. Windows HPC Server can handle a power user, a small workgroup, or a large department’s computing needs without any problems.
- Let’s look at how it works, and I will personally explain in this 10 min Youtube video, which walks you through a simple example of rendering a movie using the Parametric sweep model on Azure, I would LOVE to get feedback from you about the paper.
- Paper: Windows HPC with Burst to Windows Azure: Application Models and Data Considerations
Last but not least, I’ll have a series of exciting, already working real world demos coming up each week, whether you want to run molecular dynamics, predict the weather for the next few days, or do computational fluid dynamics design optimization for your F1 race car, I’ve got a HOWTO for you.