Jan Stephan

NVIDIA often advises that 70% - 80% of the theoretical bandwidth are already quite good. The benchmark here achieves 90%. I'm not sure how worthwhile this endeavour would be and...

We wanted to have a specialized memory allocation library outside of alpaka for a while now. Does that fit what you are envisioning for cria? I like the name. Regarding...

> Currently, we have a hard time deciding what should be part of alpaka and what should be in an external library. My two cents: * If the functionality can...

I don't think so (but maybe I'm wrong). But I had an idea for an alternative approach based on source file properties which might allow us to get rid of...

This is harder than I thought. It would be a lot easier if we could just hide `__device__` variables inside a struct :-(

I was mainly thinking about CUDA here. AFAIK there is no way to do this: ```c++ struct Foo { __device__ float f; }; ``` My initial idea to solving this...

But then we will need CI jobs for `alpaka_DEBUG=3`...

I'm putting this into the `alpaka-next` release. Every fix I make just produces two more issues and warnings. I'm honestly not sure if we should even attempt to fix this...

Okay, the checks are looking good. Have you executed the runtime tests?

Converting to draft until the performance analysis is ready.