By Anthony J. Lockwood
Dear Desktop Engineering Reader:
Do not be fooled by the title of this on-demand webinar into believing it’s just for Abaqus users. This webinar is for every CAE/CAD user out there laboring on a workstation and considering, curious about or hoping to transition from a workstation environment to an HPC (high-performance computing) environment. If you represent a small, mid-sized or even large organization, if you’re hemorrhaging money in a DIY HPC project or if you are a workstation-bound Abaqus user, this webinar is simply a must-see.
So much happens in this hour-long webinar. The penny tour is that “IBM Application Ready Solution for Abaqus Users” makes the case for an HPC setup optimized for Abaqus. The reasoning is much the same as those you know well for applications-optimized desktop workstations. Likewise, the hurdles addressed are familiar: An HPC environment is difficult to setup, manage and use. Each hurdle is thoroughly quashed.
First up is a look at Abaqus and the big view of the competitive engineering landscape — tight budgets, design and regulatory complexities, time constraints and so forth. The realistic simulation capabilities provided by Dassault Systemes SIMULIA’s Abaqus CAE system can help you bring better designs to market. But your workstation horsepower can cramp your simulation endeavors by taking too long to do complex runs, which forces design compromises, minimizes the number of design iterations and undermines your capacity to adapt to real-world pressures quickly.
This segues us to the IBM Application Ready Solution for Abaqus. Basically, this means an HPC setup optimized with the horsepower you need to evaluate more design options and more complex designs in less time. The key is that IBM designs the HPC system to make deploying, managing and using it less complex, less IT-intensive and less user unfriendly, especially when compared with a build your own cluster scheme. The aim here is to deliver an HPC system that’s as close as possible to being ready to go to work when plugged in.
While there are sample IBM Application Ready Solution for Abaqus configurations, there is no such thing as a standard configuration per se. Rather, the HPC hardware and the software are tailored for your workload instead of imposed on it. That could mean a system for workloads of four simultaneous jobs with 2-5 million degrees of freedom (MDOF) each or 15 simultaneous jobs with 10-20 MDOF each. And that could mean a “reference architecture” where high-end workstations feed jobs to the HPC cluster or engineers on thin clients work Abaqus on a cluster.
The discussions provide a quick overview of the hardware and software optimized for Abaqus. These include IBM Flex System and NeXtScale servers, storage options, and cluster management software such as IBM Platform HPC, a comprehensive workload management system engineered to take the complexity out of managing and using an HPC system.
Just before the Q&A, the presentation turns to what an Application Ready Solution can mean for you. Topics addressed here include pre-integration and testing of the hardware and software, pre-integrated Abaqus templates, total cost of ownership and system and software support. Through one of many surveys held during the presentation, users gave maximum value to a complete and integrated system and to Abaqus optimization. They gave little value to system and support, but, through bitter DIY experiences, I’m guessing that the numbers for a complete and integrated system reflect the seared sentiments of the “build an HPC yourself” crowd who had no reliable support to turn to.
The webinar’s main presentations take just 30 minutes, but stick around for the final half hour for some high-end Q&A. There’s also a link so that you can download the slides as well as a couple of PDFs with in-depth details some of the more technical aspects.
In sum, “IBM Application Ready Solution for Abaqus” makes one of the most compelling case for HPC technical environment I have seen. Very well done. Highly recommended.
Thanks, Pal. – Lockwood
Anthony J. Lockwood
Editor at Large, Desktop Engineering