Himanshu Misra, segment manager, Engineering Analysis, Silicon Graphics (SGI; www.sgi.com; Mountain View, CA), makes an interesting observation about the rationalization of utilizing computer clusters for computer-aided engineering (CAE): It’s not just the size that’s important. Sure, having a lot of processing power is valuable but, Misra points out, it’s how the hardware and the software are integrated to work most effectively and efficiently that really matters. He suggests that the so-called “white-box” clusters may be popular due to their apparent economy (i.e., the combination of a whole bunch of standard processors has a comparatively low investment cost), there are post-purchase issues including the potential of no support, as well as the issue of how effective the various programs are run on the clusters. Perhaps the architecture isn’t all that it should be. Consequently, SGI, which does offer hardware and, yes, clusters, has engineered its new offerings to the market so that they are engineered taking the CAE applications to be run into account, thereby providing a maximization of the IT investment.
SGI has just launched the Altrix XE x86-64 clusters that employ Intel Dual-Core Xeon processors and run in the Linux OS environment. These clusters are said to be tailored for improving engineering workflow productivity. Among the CAE products that the new clusters are optimized to run include LS-DYNA from LSTC (www.lstc.com); MD Nastran from MSC.Software (www.mscsoftware.com); ABAQUS FEA software from ABAQUS (www.abaqus.com); and STAR-CD and STAR-CCM+ CFD from CD-adapco (www.cd-adapco.com).—GSV