I am currently studying the usability of Unreal Engine for an intensive computing project, and Google was not very useful.
I need to do heavy computing amid the game loop for a project that is currently being implemented using OpenMP. Fluid modeling.
Unreal Engine on Windows compiles and creates using visual studio and should support what Visual Studio compiler has, but I don’t understand if you can compile the UE with the proper settings for OpenMP. Alternatively, I could easily do this work with INTEL streams using TBB blocks, which apparently are already in use in the UE ... although maybe only part of the memory allocation?
In general, I think my question is what support exists for cross-platform parallelized computing with UEs. If you need to perform a large number of identical calculations on all available kernels, how do you do it in the UE? I'm not interested in manually creating a number of threads, running them on a block of code, waiting for completion, joining, and then moving on. I would prefer a parallel_for loop for my purpose.
Moreover, I am wondering how easy it is to use other frameworks, such as CUDA or OpenCL, from the UE "script".
Unity, # dll, ++, , , #, ++... , ++, ... , , , , , . , "parallel" .