XANSYS Message: 4171 [Go back to message list] [bookmark on del.icio.us]
No rating yet Subject: Re: platforms Author: Paul Tomaszewski Date: 1999-08-24 15:33:00Thanks to all respondents [Rick, Bryan, Chris, Fern, etc]. I could tell by the length and content of the messages that you gave it some real consideration, and I appreciate you time and insight.
Bryan@b...; While I am currently on a UNIX platform, the 32-bit memory limit would not currently "cramp my style". To Chris' point [made by you as well], some intelligent finesse in model building can go a long way in reducing overall db size. Do you suppose the dual Octane handles multiple jobs better than dual PIII due to memory bandwidth? Yes, I do need Windows stuff. I still use SoftWindows on my Octane to a slight degree, but have a PC dedicated to "office" stuff. Re: the "how much money do you have?" question, not all THAT much... but the company appears willing to chip in...
Rick@c...; Good pegging of the unanswerable questions! Our company is not large enough, measured in analysts, to get a feel for the answers. And to your point about graphics, I had a bit of a time convincing IS [or IM or IT] that the graphics on the standard R&D box would just not cut it. Yes, it does weigh heavily, as does [obviously] floating-point performance.
Fern@s...; Yes, I've been watching the recent flurry of Linux news, but not ready to make the step since support is not developed. I'll assume from your medium [<100,000 DOF] vs. large [1,000,000+ DOF] distinction that above 100k the Wintel boxes start to break a sweat.
Chris@s...; Regarding your point about the ratio of solution time to overall time, I have had to make the point in the past that buying a 2X computer does not necessarily equate to twice as many jobs being done; prep and post will probably not see the same 2X gain. If my model runs in two hours, it does NOT necessarily mean that I will have two models completed in four hours. It means I will [hopefully] have one model completed and be well on my way toward an intelligent definition of a subsequent one, be it a correction or additional. And I share your belief "that faster machinery actually ends up as justification for bigger problems and more complicated methodology". If faster machinery simply allows, e.g., one to mesh automatically with tets instead of dividing into map-meshable regions, then that could be considered real gain. But if the complexity itself is increased to the level that you can no longer follow the physics of the problem [and I know I'm preaching to the choir here] with any insight, is there gain? I like the recently reiterated comment originally made by one of our British colleagues, that a model incapable of being validated must be considered invalid. But I know I'm getting way off topic here...
I think I've made all the comments I'd intended. Thanks again.
Paul R. Tomaszewski, P.E. Staff Engineer, Biomechanical Analysis
l DePuy Orthopaedics, Inc. a Johnson & Johnson company 700 Orthopaedic Dr., Warsaw, IN 46581-0988 Phone: (219) 372-7248 Fax: (219) 372-7128 E-mail: ptomasze@d...