Hi @wouterverhelst

This error can appear with simulations having a non-optimal grid with at least one very small cell leading to a very small step size. Could you please inspect the grid in your simulation and see if it can be improved? Please use "Connectivity" option in voxel view to make sure that the source is connected. If possible, please share more info, for example, parts of the log file where you can see min grid step, time step and initialization of source(s):

[INFO]: Number of cells: … [INFO]: Number of cells including PML: … [INFO]: X: Range [.. ... .. ] with minimal .. and maximal step .. [m] [INFO]: Y: Range [.. ... .. ] with minimal .. and maximal step .. [m] [INFO]: Y: Range [.. ... .. ] with minimal .. and maximal step .. [m] … [INFO]: Simulation Time Step: … sec … [INFO]: Sources: [INFO]: Initializing edge source Lines 1 (Sources) … …

Dear Habib,

Works!

Thank you!

Hello Rehmanuzair,
I think the problem is about the system connection
You have to go to task manager then you find the networking where you connect the system.
and you will be able to Run.

Could you paste the whole log?

If the problem runs in 1 GPU, could you test that?

Can you try reducing the grid resolution and see if it runs successfully then?

Hi,
"bad allocation" usually refers to errors due to not enough memory. GPUs have limited on-board memory and if the simulation size exceeds this, it fails to run. I would advice to check your system memory status, and if the simulation is too heavy, using a coarse grid (less MCells) can help.

@Mélina This is close to what I want but not exactly. I still want ARES to say I can use both GPU's but I want to be able to say "for simulation 1 run on GPU0, for simulation 2 run on GPU1" . Is there anyway to do this? I tried messing around with the concurrent jobs number (setting it to 2 or 3 or 1) in ARES but I don't believe it's working correctly as I still am only running 1 simulation at a time in the queue.

Try and do it programmatically in the python scripter. Set one figure up how you want it then right click on the final slice viewer in the explorer tab and click "to python". Then just look for the bounds there, save the script, and use that for all of your other figures.

Yes, bad alloc means the memory could not be allocated. Try to reduce grid size, and ensure your subgrid is substantially smaller than the total computational grid.

Thank you Ofli for your quick response. That was very helpful!

Dear Liu,
Yes, in order to see any losses, you need to set at least a conductivity different to 0. I would advice you to check if you set correct electric properties in the material settings.
Best

Hello,
Do you have any error message appearing in the console?
If a grid cannot be created, you have to carefully check the grid settings, and ensure that the number of cells is not the big (if so, you can reduce it by adjusting the padding, for example).

Dear @habib,

Just to inform you that has been solved the bug which was causing this pop-up window to appear, and which stopped the script from continuing until the "Ok" button was pressed:

bug.JPG

However, this bug was caused by the code line, which appears in the image:

"analysis.RefreshViewers()"

With the new S4L v7.0, this line caused this bug for some reason. After removing it, it was possible to avoid that pop-up window. Just to highlight that finding this bug costed 10+ hours, due to the new update.

Best,
Guillem

@ofli Thank you for texting.
Yes. I set up the similar one. I adjusted the size to the edge of the TEMcell tho.