Dear Desktop Engineering Reader:
Let me flop this fish right up front on the table: I love the idea of high-performance computing (HPC) environments for engineering. On the one hand, engineers need the kind of power that an HPC cluster gives them to bash out some multiphysics, CFD, or other monster analysis. On the other hand, I really hate the idea of engineers having to play IT guru to leverage a cluster. Engineers should not have to be expert in putting together and using clusters. On the third hand, I know that many small- and mid-sized businesses cannot afford HPC and are sitting on the sidelines. In my opinion, this is no time to be turning away clients because you are underpowered.
You may recall that a couple of weeks ago I wrote you about Cray's CX1 deskside personal supercomputer. The CX1 has a lot of characteristics that I really like, but two really spark my HPC interest. First, the CX1 is designed and optimized to be as easy to set up and use as a personal computer. Second, it's an HPC cluster resource that has been designed and optimized for engineering outfits where HPC has been an unaffordable option. You would probably spend as much dough customizing a workstation or two as you would on a CX1 that gets you into HPC with more power.
The CX1 is certified to be Intel Cluster Ready and can run Red Hat or Windows Server, so it can slip into your current environment and change the way you do businesses in a hurry. In practice, that means the CX1 provides you with the ability to not only run scalable, high-end applications like ANSYS, but it dramatically reduces the amount of time it takes to arrive at a solution as compared to your workstation. By “dramatically,” I mean a solution in 30 minutes that once took you three hours to get.
But, don't believe my blather. See for yourself. Cray and ANSYS have teamed up for a 26-minute on-demand webcast that both introduces you to CX1 personal supercomputer and shows you what ANSYS FLUENT and ANSYS Mechanical users can gain with the CX1. The ANSYS discussions offer some pretty impressive simulation benchmarks that you'll want to check out to get a better idea of what I meant by reduced time to a solution. Map it to your work. I liked the results of a 14-million cell mesh FLUENT benchmark using different numbers of cores.
I saw my first Cray supercomputer in an air-conditioned static-free room back in the early 1980s when I was writing about PCs with 256K of RAM and dual—dual!—360K disk drives. Suffice to say, it was one of those life moments where you hear the “Hallelujah" chorus in your head. “This is computing as it should be—Ahhhhh.” In 2009, affordable HPC kind of has that effect on me. Check out the video and see if you agree.
Thanks, Pal — Lockwood
Anthony J. Lockwood
Editor at Large, Desktop Engineering Magazine