how to use double precision I model the case where the grid size difference is large. the grid usage is like compressible flow grid usage. so how can I do when simulate such kind of grid. someone says it is way to use the double precision. but after setting the double precision,the case can not run.how can I do ? help

star -dp

If -dp cannot run, normally it is due to memory (RAM) problem.

I don't know whether the Ram is enough(use 2G Ram)but my total cell number is about 2005000.

If you use Window XP machine, it is marginal. Did the virtual memory not enough message pop up at system tray/task bar?

What other features do you have in the simulation? Stuff like lagrangian etc really eats memory, make sure you don't have any other software running, especially memory hungry stuff like pro-STAR or any other CFD/FEA packages. Also if you are using window$ hit control+alt+delete, look in the task manager and see what the memory usage is like of if using linux, just type "top" in a new shell.

What happens when it doesn't run, does it start up and crash, crash a few iterations in .....be more specific please.

Hi Ben,

How many percent the deviation between sp and dp?

There isn't really a way of saying dp is x% more accurate than sp it is somewhat case dependend. Douple precision variables can contain 17 digits before truncation where as single precision ones contain 10 therefore you get more accuracy and less rounding error in dp compared to sp. Certain applications dp is a must (l2p for example) others it is less important and is just a waste of resource. Run some tests yourself and compare...

Thanks Ben