How to deal with calculation results of indefinite length in cuda?

In my kernel function, each thread performs the same calculation process, but may produce a different number of calculation results.
For example, the calculation result of thread 1 contains 5 integers, while the calculation result of thread 2 will contain 2 integers, and the calculation result of thread n will contain 10 integers, how can I deal with this situation?
If you can provide a solution or related information, I will be very grateful.

There are a number of questions on various forums that address this topic. Here is one.

Ok, thanks for your help!

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.