ILGPU Help optimizing my code

I am trying to use ILGPU to make my code execution faster. The way I did it, it is around 60% faster when I run it with a cuda accelerator, but when I run it with a CPU accelrator it is much slower than my normal CPU approach without ILGPU.

Here is the code
I suspect all these allocations make my code much slower than it could be. I would appreciate any suggestions to make it faster.
Thank you
no comments yet
Be the first to share what you think!
C# devs
null reference exceptions

source