Ivar KJELBERG
COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Jul 25, 2012, 2:17 a.m. EDT
Hi
I'm on travel forsome time and have no access to my COMSOL WS so I cannot open your model, but 3D models do require quite a lot of RAM, once they become slightly complex. And 2Gb, once you give away 2/3 for your OS, is not really "much" or a FEM solving programme.
Have you considered using symmetry, or dimension reduction (2D-axi , 2D or other if it applies) ?
--
Good luck
Ivar
Hi
I'm on travel forsome time and have no access to my COMSOL WS so I cannot open your model, but 3D models do require quite a lot of RAM, once they become slightly complex. And 2Gb, once you give away 2/3 for your OS, is not really "much" or a FEM solving programme.
Have you considered using symmetry, or dimension reduction (2D-axi , 2D or other if it applies) ?
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Jul 25, 2012, 3:07 a.m. EDT
Hi,
I could not open your model because I use an older version of COMSOL.
2 GB RAM is usually not enough for 3D problems and you need to use symmetry or make a 2D model instead. Because using a coarse mesh is not very good idea. I can recommand you that when you use COMSOL on a PC like that close all the other programs like email software or any program which also uses memory but even after doing this you probably still receive this error message
Hi,
I could not open your model because I use an older version of COMSOL.
2 GB RAM is usually not enough for 3D problems and you need to use symmetry or make a 2D model instead. Because using a coarse mesh is not very good idea. I can recommand you that when you use COMSOL on a PC like that close all the other programs like email software or any program which also uses memory but even after doing this you probably still receive this error message
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Jul 25, 2012, 3:59 a.m. EDT
Hi Ivar,
I am not sure if I could use 2D model, because I used to compare with actual experiment which has 3D structure.
Could I apply 2D model to 3D? I f so, how could I do that?
And about symmetry, I would like to try that.
Would you explain me a little bit about how to do symmetry?
I am sorry that my questions may be very very basic..
Hi Ivar,
I am not sure if I could use 2D model, because I used to compare with actual experiment which has 3D structure.
Could I apply 2D model to 3D? I f so, how could I do that?
And about symmetry, I would like to try that.
Would you explain me a little bit about how to do symmetry?
I am sorry that my questions may be very very basic..
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Jul 25, 2012, 7:55 a.m. EDT
Hi,
Another solution to the memory problems is to use ubuntu as the operating system. Ubuntu uses memory more efficiently.
Atilla
Hi,
Another solution to the memory problems is to use ubuntu as the operating system. Ubuntu uses memory more efficiently.
Atilla
Ivar KJELBERG
COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Jul 25, 2012, 4:13 p.m. EDT
Hi
It's true that Linux handles somewhat better the RAM andcan be configured to use less than other OS, but the difference is not that bi, and if you have only 2Gb, you will swap rapidly and then any OS is sloooow.
A 2D model is never fully "as precise" as a 3D but it can often be a good approximation and as you use less RAM you can use a denser mesh and hence obtain intersting results, even if 1 dimension is missing (or assumed infinite and constant). But again, it does not work for all models, it depends strongly on your geoemtry too
--
Good luck
Ivar
Hi
It's true that Linux handles somewhat better the RAM andcan be configured to use less than other OS, but the difference is not that bi, and if you have only 2Gb, you will swap rapidly and then any OS is sloooow.
A 2D model is never fully "as precise" as a 3D but it can often be a good approximation and as you use less RAM you can use a denser mesh and hence obtain intersting results, even if 1 dimension is missing (or assumed infinite and constant). But again, it does not work for all models, it depends strongly on your geoemtry too
--
Good luck
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Jul 25, 2012, 8:19 p.m. EDT
I think I should upgrade my RAM then if I want to do more difficult structure..
What I am not clear yet is to do symmetric mesh. My structure has 1/2 symmetry so I would like to try symmetric mesh..
What step should I take to do that?
Thank you
==
Rheza
I think I should upgrade my RAM then if I want to do more difficult structure..
What I am not clear yet is to do symmetric mesh. My structure has 1/2 symmetry so I would like to try symmetric mesh..
What step should I take to do that?
Thank you
==
Rheza
Ivar KJELBERG
COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Jul 26, 2012, 4:06 a.m. EDT
Hi
Certainly a lot of RAM is usefull for FEM, but you need also an OS that can handle more RAM (>2Gb), which means a 64 bit compatible OS. Then its the hardware, the bandwidth CPU to RAM is also of importance, at some stage you start to loose time as you might spend more time shuffling data between Cash and RAM and CPU than really number crunching, it all goes together
--
Good luck
Ivar
Hi
Certainly a lot of RAM is usefull for FEM, but you need also an OS that can handle more RAM (>2Gb), which means a 64 bit compatible OS. Then its the hardware, the bandwidth CPU to RAM is also of importance, at some stage you start to loose time as you might spend more time shuffling data between Cash and RAM and CPU than really number crunching, it all goes together
--
Good luck
Ivar