Recently I’ve started to do GPGPU programming and I’ve stuck with the problem of dynamic data allocation. In my CPU code I use lists to dynamically remove and add elements to it. But in GPU programming list data structure is not supported as the buffer size is constant during the dispatch. So, my concern is whether there are any solutions to my problem. What algorithm can I use to replace dynamic data allocation in GPU code?

Thank you in advance!

P.S. I’m using HLSL for GPGPU programming.

Anonymous Asked question May 14, 2021