moldflow should be able to estimate the time for analysis

[madhukeshwart]

moldflow should be able to estimate the time for analysis.

when ever, we are launch an analysis, moldflow should be able to pop up tentative time to complete the analysis.

it should be based RAM/ memory / processor availble/solver settings and number of nodes/elements

[raalteh]

Although I understand the request, it's nearly impossible to do with a reasonable degree of accuracy ahead of time for various reasons:

1) Analysis time is not necessarily a straight function of the number of elements.You can check this for a very simple case (for flow for instance) if you model a very long flat plate, and gate it at the end, vs. gate it in the middle. You will most likely see a different analysis time.

2) It's impossible to predict ahead of time how quick convergence will be reached for any given timestep.

3) you don't know ahead of time what else is running on the machine that take memory and CPU.

I indersand that giving a ball park would be great already, but we would give a very rough at best. But if we can't predict it with sufficient confidence (ahead of time, or time remaining), it doesn't have much more value than your own experienced based estimate of 'this will take about an hour' IMHO

[raalteh]

I would like to have other people's perspective on this; if this is a significant amount of work involved (that we would for instance not dedicate to for instance accuracy) and the time estimate would be a very very rough ball park, it this worth doing?

[mayur_dhumal]

Hi Hanno,

I am afraid it might be POSSIBLE with present version of Synergy as well, you will just need to create a VB script, which can be played as macro within Synergy.

Factors to be considered while measuring ~appr. time will depend upon:

Available System RAM during solving / Memory / Paging Size / Number of Cores / Number of Elements (Tria/Tetra) / GPU availability / Parallization / No. of threads etc. info.

Regards,

Mayur