How to Optimize Performance for Handling Large Data with Archicad's C++ API

[nishida_jp]

When handling large amounts of data with Archicad's C++ API, the processing time is too long. What techniques can be used to process data more quickly?

For example, is multitasking possible? Please share any effective methods.

[Ralph Wessel]

It's possible to run concurrent processes in an add-on, but with significant restrictions. My general advice is:

• Never call any API functions from any thread apart from the one the add-on is invoked on, i.e. don't use the API functionality concurrently. I can't rule out some functions being safe (only GS could advise on that) but assume they are not thread-safe
• Don't try to run a process independent of invocation from Archicad, i.e. don't have something continuously as a child process of Archicad but running independently.

Some possible uses might include data-crunching once the requisite data has been obtained from the API - you could break the data into chunks and fire off multiple threads to process the chunks in parallel. But this would be started by a call from Archicad into the add-on, and this call should not return until all the threads are finished.

Take this as general advice rather than a completely definitive answer - GS might be able to provide some exceptions. But I expect this will generally set you on the right path.