AnsweredAssumed Answered

How does a quad core help SW performance?

Question asked by Mark Matthews on Jul 22, 2010
Latest reply on Jul 26, 2010 by Mauricio Martinez-Saez

With all these new chipsets coming out from Intel like the core i5's and i7's, it seems like any relatively high end machine our office is going to get (I'm talking laptops for senior engineers who may need to run SW on them in a client meeting; not as the primary workstation) is going to have a quadcore processor in it. Now I understand that SW must process a model squentially to rebuild the model tree, but can it multithread some operations as it's rebuilding a model. For instance, if it's up to calculating a fillet, will it use all four cores to do that; then move on to the next thing? I've tried testing this by watching my performance graph while attempting a large pattern fill. All four cores start processing the pattern at 25% CPU usage, then one of them started doing most of the work and total CPU usage whent to 45%, and then the're all about even and running at 25%. It took a total of 8 minutes to calculate. So this means SW can use all cores available to do some of the heavy geometry creation? Why wern't all the cores pegged at 100%. Hopefully there is someone here that knows this stuff and can help clarify this question for me?



I've just run a rebuild on the test model and set the SW affinity under the "processes" tab to have SW run on only one core. That core pegged out at 100% during the rebuild and the feature statistics stated the fill pattern took the exact same time ot process as with all four cores working. I am now confused.