Solver uses swap / to less RAM / large grid

matmax-168 · April 14, 2015, 06:41

Hello,
I try to simulate a flow around a wing (ansys cfx). I have chosen a huge control volume to determine induced velocities far behind the wing. When I refine the mesh (~4M knodes unstructured) the solver output is saying "Increase the integer stack memory size"....so I increased the Memory Allocation Factor, but this did not made the Solver run.

Second step was to try a coarser grid (~2,5M knodes unstructured) and the solver started, but after more loops or immediately he uses swap.
Is it still possible to run the simulation with fine grids even with the available power of my workstation? (for the cost of higher calculation duration)

4x Intel (R) Core (TM)i5-4590 CPU @ 3.30 GHz
8Gb Ram

Any help is thankfully welcome!

ghorrocks · April 15, 2015, 06:25

If it uses swap on the coarse grid then you will certainly run out of memory for the fine one. If you want to run the fine mesh (or the coarse mesh efficiently without swapping) you need more memory or to run distributed parallel.

April 14, 2015, 06:41	Solver uses swap / to less RAM / large grid	#1
matmax-168 New Member Join Date: Dec 2014 Posts: 9 Rep Power: 11	Hello, I try to simulate a flow around a wing (ansys cfx). I have chosen a huge control volume to determine induced velocities far behind the wing. When I refine the mesh (~4M knodes unstructured) the solver output is saying "Increase the integer stack memory size"....so I increased the Memory Allocation Factor, but this did not made the Solver run. Second step was to try a coarser grid (~2,5M knodes unstructured) and the solver started, but after more loops or immediately he uses swap. Is it still possible to run the simulation with fine grids even with the available power of my workstation? (for the cost of higher calculation duration) 4x Intel (R) Core (TM)i5-4590 CPU @ 3.30 GHz 8Gb Ram Any help is thankfully welcome!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
elliptical solver - grid lines going into the surface	syoo	Pointwise & Gridgen	5	January 23, 2014 17:05
Stretch Grid Poisson Solver	G. Wilson	Main CFD Forum	0	February 19, 2002 12:16
matrix solver about block structured grid	young min, Sohn	Main CFD Forum	1	July 11, 2000 19:35
solver for linear system with large sparse matrix	Yangang Bao	Main CFD Forum	1	October 25, 1999 04:22
Incompressible flow solver (staggered grid)	J. Ehrhard	Main CFD Forum	1	October 8, 1998 19:47

April 15, 2015, 06:25		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,700 Rep Power: 143	If it uses swap on the coarse grid then you will certainly run out of memory for the fine one. If you want to run the fine mesh (or the coarse mesh efficiently without swapping) you need more memory or to run distributed parallel.