i don't understand the workflow here
you're saying ram increases over long periods of using solidworks?
or you're saying that ram increases after you hit run?
does this problem only happen with this model?
The workflow is the following:
Open Windows: RAM memory 3.5 GB
Open Solidworks: RAM memory 4.27 GB
Open the model: RAM memory 4.72 GB
Open the SW Simulation add-in and non-linear study: RAM memory 4.78 GB
I save the model with a different name, memory goes up to 5.15 GB and then down as previously 4.83 GB.Until now everything is fine, but as soon as I go to the model and modify something in the scketch the memory stacks up.
It seems to me that at every rebuilt, Solidworks is calculating the new geometry and ocuppy an amount of memory space for that, which is OK.
Here is an example of memory stack up after rebuilt, te memory raised to 8.84 GB:
After the calculation is performed and the model is updated, it is not OK to keep memory ocuppied, Solidworks should dump the memory to make room for other calculations. After a couple of modifications and rebuilts in the model, the memory goes up to the ceiling and then freezes or crashes. So to avoid that I need to constantly monitor the memory stack up and to force shut down Solidworks to dump all the loaded memory .
This happens in a time span of 5 or 10 minutes which is the approximate time span of the Windows Task Manager in the above picture.
Of course it depends on the compexity of the model and the complexity of the calculations.
In my first post the memory stack up is done by NSTAR when solving the study, but in my second post the memory stack up is done by Solidworks, so I suspect there is a bug in the memory management of Solidworks. In my opinion it happens in all models, at least I saw it in all my models, although I build them in a similar way using the circular pattern feature.
I think this bug is not visible in simple models with few features. Actually in my model which is a gear wheel I have just 3 Boss-Extrudes and 3 concentric Circular Patterns, but because one of the pattern is a 600 elements array and the other two patterns 300 elements arrays, the model rebuild requires a big amount of calculations. Because the model has a large number of array elements, this also implies that my simulation mesh has a large number of DOF, anyways I have more than 200000 DOFs in all my models. But as I said after the calculations are done and the new geometry solution is found, the previous data in the memory should be dumped to be replaced with the new data. I don't understand why after each calculation the data is kept in memory and never dumped.
Why there is not a function in Solidworks that verifies the amount of free memory in RAM and when is critically low, it saves the data on the disk, in the case this data is really necessary to be there for Solidworks calculations? Is this not a common practice in the CAE/FEA world?
post your model for confirmation or report to your reseller to get it confirmed
i'd suggest checking on another computer before you do that as this may be an install/hardware issue but i think you're right that there may be a software issue that is either general or specific to your model
I reported to the reseller a while back but they too requested my model.
Unfortunately I cannot provide the model for confidentiality reasons.
Also I tested on different computers and the behavior is the same.
But I can provide some clues to build a similar model.
if you build it, they will test.
OK, I will do that.
was your reseller able to confirm the issue?
I didn't had the time to open a ticket with them, but in the mean time I saw inconsistent behaviour. It doesn't happen all the time it depends on the models. My old models built in SW2013 and run in SW2015 they stack up the memory but then I build the same model from scratch in 2015 and did the same but only if I used Intel Direct Solver, with the older Direct Solver the memory doesn't stack up, which is ok the only thing with the old solver is that is using only 4-5 cores out of 12. The Large Direct Solver does not work at all. And still, if I go to the model and modify the sketch then during rebuilt the memory stacks up and never dump the garbage. When I'll have some spare time I'll contact my resellers. Thank you for checking me up.
there is no cap on the number of cores used by the old direct sparse solver, it will use all of them if it can
it sounds like you have 2 issues
1. when using the direct sparse solver, the ram is not being dumped when an operation is complete
2. when rebuilding, ram is continually used (memory leak)
if these are happening on any model, this would be mode widespread, this may be a system specific issue
again if you want to post a model, I can check it on my system
I have this same problem
Memory build up over day while opening and closing Parts and Assems till you get up to bout 13 GB of memory then starts crashing
Even after closing solidworks completely it was still up around 13GB
Doesent free up until end processes of SLDWORKS.EXE
which is still running in processes but not applications
Nvidia Quadro K2000