How to speed-up GUI for large structures?


#1

We are working in a model comprising 33x33x21 spheres. FDTD GUI works fine in constructing the structure and meshing it, it takes quite a long time, but this is ok (the command redrawoff, helps). The problem is that this structure slows done each and every modification we make in the model (Say for example, adding a source or a monitor). A way to speed-up the design procedure is by disabling the group of spheres. However, once we enable it again, everything goes very slow. We wonder whether exists some sort of “draft” mode, where the group is represented graphically by just a box or something similar.


Initiating structure groups with a very large number of objects
Speed up GUI when making large number of objects
Xyビューウィンドウをスクリプトから閉じる方法
Disabling the GUI Components of FDTD during MATLAB Optimization
Overlapped truncated cones in FDTD
GDSII export error
How to close XY view via script command?
#2

If you go to the ‘edit properties’ window of any geometric object, under the ‘Graphical rendering’ tab you will see an option called ‘render type.’ For large number or complex structures, you can reduce the burden on the GUI by choosing the ‘wireframe’ option (instead of the default ‘detailed’ option) so that the structure is represented by a wire frame only. This should reduce the burden on the GUI and speed up your work environment. The only catch is that you will have to turn the ‘wireframe’ option for each sphere separately. However, your should be able to do this easily from the script.


Adding lots of dipoles to simulation
#3

A couple of additional suggestions are to reduce the detail setting which can also be set in the “Graphical rendering” tab of the object editor to speed up the drawing time, and close unnecessary CAD view ports. For example, if you only need to see the XY view of the object, you can close the other views. You can always re-open the view ports later from the View menu in the main title bar.


#4

Thanks for the tip. We tried the wireframe setting, but didn’t notice a significant improvement.


#5

This is something curious, because even with all the views closed we had to wait for the GUI to respond. I forgot to mention that we were working in remote, meaning through an x-server. Each time we made a modification, a huge network traffic started. Perhaps in local everything goes smoother. I will make a few trials and come back to see whether we can identify were the bottleneck is.

Thanks for your help.


#6

Hi @jose_llorens. I am not an expert on how the remote connection should affect your GUI’s response. However, in my personal experience, I always sense a little lag in the response of the display while accessing another computer using remote desktop. This probably has to do with a large amount of graphics data passing through the network. May be that is what is affecting your GUI response as well?


Simulation of inverse opal
#7

I’d gone through similar GUI slowdown due to the large number of geometries or array just like @jose_llorens. Recently, I’d built 275X375 (~100,000) spheres (green color) in my simulation model as follow.

To avoid GUI slow-down, at first you can add one sphere by suppressing the detail function in the Graphical rendering tab as below.

Then, you can make it a group (Add to new group by right click) to treat the whole array objects as a single object. In the group you can duplicate many objects using Array function. Now even though there still exists inevitable slowdown occurrence, you might experience the enhancement of GUI speed in the simulation model, compared to when the rendering detail option is not adjusted. You’d better finish the large-scale array at the end of your optical modeling procedure.
The GUI response depends on the specifications of graphics card, CPU, and memory of your simulation machine. I guess the capability of graphic card is quite sensitive to the GUI performance when a large number of objects is implemented in the model. Here is the some specifications of CPU and graphic card of workstation for my optical simulation.

  • Xeon® CPU E5-2687W v3 @ 3.1Ghz (2 CPUs) 256Gb RAM
  • NVIDIA GeForce GT 740 (GPU: 384 cores, 993MHz / Memory 1.8Gbps, 2048 MB) (link for detailed specs)

Efficient packing of non-overlapping Spheres and Shells with radii drown from Gaussian distribution with a certain mean and standard deviation