René Widera

I Opened again a ticket for this issue. I think to remember that @SimeonEhrig already did this a year ago but I assume it was ignored. Let's wait for the...

@fwyzard To let the CI pass ypu must update the filter and comment https://github.com/alpaka-group/alpaka/blob/e9dde8585e8584b1512c7e8c6bb0ecf0f3a5151d/script/job_generator/alpaka_filter.py#L139-L143 to ``` if row_check_version(row, HOST_COMPILER, "==", "20") and row_check_version( row, DEVICE_COMPILER, "

It is not only the mdspan test which is failing. ``` Start 47: memViewTest 47/55 Test #47: memViewTest ......................***Failed 1.31 sec Randomness seeded to: 2720421228 /builds/hzdr/crp/alpaka/include/alpaka/mem/buf/uniformCudaHip/traits/BufUniformCudaHipRtTraits.hpp(252) 'TApi::free(ptr)' returned error :...

@fwyzard I can reproduce the error and the error is gone if you update the fetched mdspan version to 0.6.0. I am not allowed to push to your branch therefore,...

@AuroraPerego If this PR should go into the release 2.1 you should get the CI green until the latest 11.12.2025 that we have at least one day to prepare the...

> > @AuroraPerego If this PR should go into the release 2.1 you should get the CI green until the latest 11.12.2025 that we have at least one day to...

Dynamic shared memory follows the cuda principles of dynamic shared memory. You need to define the amount at runtime outside of the kernel and within a kernel you are responsible...

That is a very good question, from a short look it looked wrong and `ALPAKA_FN_HOST` should be fine.

Thanks for openeing the issue. This is indeed something important. I invested already a lot of time to think about it. A possible way to implement it would be having...

sry have not checked you prototype branch. You did it already via trait, nice. The question is now how we could makenit generic that it fits all accelerators, of possible.