CBDM Exposure with Radiance 6.1a: rtrace multiprocessing on Windows

[Ole Marius Svendsen]

Hi Ryan, I hope you are doing well. Long time no see.

I have been looking into the newer Radiance 6.1a release, particularly because of its improvements to multiprocessing on Windows, which seem promising for speeding up larger Radiance calculations. I was wondering whether this could also be used to improve the calculation speed of VI-Suite's CBDM Exposure calculations.

I am currently using Blender 5.0 with VI-Suite 0.7 on Windows. The addon has been working very well for me on Blender 5.0, so I have not had a reason to move to a newer Blender/VI-Suite version.

I previously had the Radiance version provided through pyradiance, which was Radiance 6.0a. I have now installed Radiance 6.1a separately and configured VI-Suite to use it through the Radiance bin/lib directory settings.

I have confirmed that VI-Suite is indeed calling Radiance 6.1a. Running rtrace -n 24 directly also works, so Radiance itself accepts multiprocessing on my 24-thread CPU.

However, when I run a CBDM Exposure calculation, I get:

rtrace: warning - only single process supported

Task Manager also shows only one rtrace.exe process during the calculation.

I found that the CBDM calculation code in vi_operators.py generates commands along the lines of:

rtrace -n {nproc} ... -I ...

so I was expecting multiple Radiance processes.

Do you know whether the CBDM Exposure implementation in VI-Suite 0.7 is intentionally limited to a single rtrace process on Windows, or whether there is a way to make it use multiple processes/cores with the newer Radiance versions?

[VI-Suite]

Hi Ole.

Yes, been a while. Trust you're well.

Off the top of my head it could be that as previous versions of Radiance restricted multi-processing on Windows for certain programs, that the VI-Suite is enforcing single core on those programs. If the newer version of Radiance has relaxed those restrictions I'll have a look to see what can be changed in the code to take advantage. I'm busy at the moment so will be a couple of weeks before I get round to it.

Cheers

Ryan

[Ole Marius Svendsen]

All good, its been a slow entry to the autumn season. Seems as two wars and higher intrest rates has slowed down the building marked substantially. Alas it gives me some time to further experiment on work flows. Helps alot with AI to write code (as i dont know programming) to make scripts preparing and IFC files window surfaces from two faces to one, and currently doing a investigation on how to detect whats inside/outside surfaces. Its a bit slow work, but i have had some progress to isolate the outher face so i can quicker use watt analysis on window surfaces to determine where one need blinds in apartment blocks, with hopes i can eventually drop DL Light and sketchup altogether.

Hopfully new semester has brought good student eager to learn, and your research is going well. 

No worry! It was a interesting potential i stumbled on, and i wanted to making a query to isolate as to what causes this. I hope by narrowing this down it will further help you as the developer.

Further, I'm also investegate how to isolate results on rhoughly 1 500 window surfaces, interestingly enough the VI grid seems to be displeased when i have active that many objects (clean surfaces) with different orientation, and no longer wants to grid the intent sizes. I am deffently stress testing the work flow! However with some AI generated scripts i was still able to cut the mesh into smaler faces if all the objects are in "object mode" instead of "edit mode". Lastly i am hoping to recreate tresshold values in my own intervalls to display them on the window surfaces. Lik ex. 1 800 - 1 500 - 1 200 - 0 W/m2. or more.