increase ram memory used by ansys 13

[kasak985]

Hello everyone!

I would like to ask you how it is possible to increase the memory that uses ansys in the calculation in workbench mode since I own a pc with 16 gb of ram but ansys 13 wb, according to task manager, uses "only" 4,7 gb at most.

Thank you! ! ! !

 

[stefano.garbin]

insert a snippet command on the "static structural" part (or equivalent).
in the command snippet write:
bcsoption,, incore,13000
so ansys will try to allocate 13gb of rams

ps: is it not that maybe ansys for that particular analysis is enough 4.7gb? It's no good for him anymore. What do you use? the direct scatter solver? check in "solution information" if your analysis is turning in "optimal in-core mode"

 

[kasak985]

I'll cover you off what I think it means to stay out-core.
I am performing a pre-stressed modal analysis with 400 thousand elements... .
This is what I extracted from structural analysis
modal analysis is still calculating from more than one hour and employs "only" 5gb of ram...

scattered matrix direct solver.
number of equations = 2591577, maximum wavefront = 6252
memory allocated for solver = 4615.311 mb
memory required for in-core = 21785.366 mb
optimal memory required for out-of-core = 1263.387 mb
minimum memory required for out-of-core = 479.551 mb

[bleep] [bleep] cp = 184.050 time= 17:39:19the sparse matrix solver is currently running in the optimal
out-of-core memory mode. this memory mode achieves a good balance of
memory usage and hard drive usage in order to obtain an efficient
solution time.
sparse solver maximum pivot= 189450367 at node 280957 uz.
sparse solver minimum pivot= 2300.72604 at node 546548 uy.
sparse solver minimum pivot in absolute value= 2300.72604 at node
546548 uy.

from workbench, clicking on solution in the structural window, I saw that the solutor is the mechanical apdl
the command you said where it fits precisely: in workbench or in mechanical? ?

Thank you for your concern. . .

 

[kasak985]

on mechanical, selecting static structural, button command.. .
Right? ?

 

[stefano.garbin]

memory required for in-core = 21785.366 mb

your analysis takes about 22gb of ram memory to turn "in-core", which is the fastest mode.
the solutor choose it by going on "analysis settings" as from attached image

 

[stefano.garbin]

on mechanical, selecting static structural, button command.. .
Right? ?

Right, right.
However from version 13 the bcsoption command in workbench should not be necessary.
note that ansys requires a memory block ram "continue" to turn: is unable to use detached blocks of ram and consider them as a single.
for this reason you will never be able to allocate to ansys all the memory that seems to you to be free.

 

[kasak985]

Thank you very much

that I know the iterative method would be even longer for calculation... should only be used in case of convergence problems...right? ?

if I can't give him more memory manually to speed up calculation in workbench I can only:

1) decrease the number of elements. . .
2) change ram or use cluster-type systems.. . .

Hello

 

[MarioMcP]

I'm just gonna hook up with this discussion.

launching an analysis shows me this error:

after banging our head for a while I changed computers and gave them no problem.

Since I'm supposed to be 4 hours of train a day to change computers, I'd like to try to fix it. Thank you in advance

 

[stefano.garbin]

are both 64-bit computers? (i.e. the version of ansys is the 64-bit compilation version).
what size has the file database? (if you use workbench searches for *.db files in the project subfolders)

 

[MarioMcP]

no, on my laptop I have the 32bit version because the computer is 32 bit (of course)

on the lab computer instead we have 64 bits.

the strange thing is that I launched the next cycle that is much more challenging and, despite taking 3 hours, it didn't give me

problems. here the error gives me when I launch I ask him to set the initial temperature to 20°c.. .

imposed the intial condition
imposed the time
I give the solver.

removing the solve, zero problems.

the file is 400mb

 

[stefano.garbin]

That's weird. Are service packs aligned?
of ansys 13 there is the original version and then the sp2 if I don't remember badly.. .

 

[MarioMcP]

I think I figured out what the problem was: Ram.

on the laboratory computer, which is 64 bit, it all works properly. In particular, immediately after the pto in which it stuck, tells me that the ram memory required is 2.5gb. I have a 4 giga, considering that windows and other programs that turn absorb from 1.5 to 1.8gb it is evident that, adding another 2.5gb, go beyond 4.

But the thing that I don't explain is that I also have 4gb in the lab and how performance are equivalent, the only different thing is that my is 32 bit while the other is 64bit

 

[gerod]

from experience, between 32 and 64 bits with ansys there are abyssal calculation times.
the 32 does not manage more than 3 gb approximately of ram. consider the quota so and you remain little for ansys.
It happens to me sometimes that the 32 bit crashes, then I move to 64 and goes smooth like oil. If you want to save time passes to 64.
Hi.

 

[stefano.garbin]

eye boys we have to specify some things.

1-between 32 and 64 bits (if the solution turns all in the ram and does not write on disk) there are no appreciable differences as resolution time.

2- 32-bit systems: theoretically manage up to 4gb rams, practically only 3gb. from 3gb it is necessary to remove the ram consumed by the operating system and other applications. attention that ansys, when launching the solution, needs to allocate contiguous memory blocks, so if the use of ram tells you that you have 2 free gb but these are "sick" in two parts, will never manage to allocate the ram.
from practical experience, if you launch ansys on a 32bit windows system without having other open applications you can allocate 1.3-1.5gb at most.

3-on 32-bit winsozz you can try to increase the available ram. see this link valid for win7 and view.

4-even worse if you are using a 32bit so laptop. Many low-end laptops share the ram dia for the processor that for the video card. therefore the usable quantity can also fall of another medium gg.

5-Don't try to increase disk swap space to make ansys believe that you have more rams. ansys swap file uses it in a disastrous way with severe performance degradation. in the case of out-of-core solutions (sparse solver) or iterative solutor let ansys manage writing on disk.

6-in the specific case that leads to crash what kind of solutor are you using? with what parameters?

 

[MarioMcP]

My laptop, when I took it three years ago, was on the other side but now it's obviously Jurassic

what I could find is that with the computer of the lab the calculation times were quicker...the same cycle done with my laptop in 1.5 hours yesterday was made in 45 minutes.. .