Hello everyone,
I recently discovered that Radiance only performs simulations on Windows with a single processor. However, according to what Sarith Subramaniam mentioned here, this limitation has been addressed in Honeybee, allowing for simulations using multiple cores.
I couldn’t find the code for this part of Honeybee accurately on GitHub. Is there any explanation, repository, or documentation available regarding this?
Thank you!
Hi @Soushiyant,
You can set the number of workers in HB Recipe Settings.
If you have 8 workers it will split the sensor grid or image into 8 equal parts and run them simultaneously. When they have finished the results will be merged into one sensor grid or one image.
For grid-based simulations there is a minimum sensor count to avoid splitting a grid into very small parts, which will add unnecessary overhead. If your sensor grid is very small you will not see any benefit, since it will still use the single processor approach for Radiance on Windows, but for a larger grid it will split it and run several single processor calls at the same time.
2 Likes
Hi dear Mikkel
Thank you for your response.
Perhaps I didn’t phrase my question well. I am looking for the method of this classification and the Python code you developed for it. The question that has arisen for me is that, for example, when we perform a simple grid-based simulation with one core using Honeybee and Radiance, it takes longer compared to when we do the same work with Radiance itself, which seems to be due to the way the Queenbee is structured. However, if we set the number of cores to 8, the difference in their speeds becomes more.
why this happend?