Damien L-G

@crtrott `atomic_nand(ptr, val)` is equivalent to `_logical-not_(atomic_and(ptr, val))`

On 2nd thought I am not convinced that we should do anything about NAND. I'd almost rather provide it than explain why we removed it.

A late addition to the list of oddities: * atomic ops lshift and rshift currently take a `T` operands but I wonder if we should change it to take an...

I forgot to say the HIP backend has the same behavior. Note that if using the 3-argument overload of deep_copy() and passing an execution space, the error would be async...

I am going to interpret the feedback so far as "not a feature". As far as the debug checks are concerned, ideally these would happen at (unmanaged) view construction but...

Do you want to open a PR to fix these?

The issue has only been partially addressed: they are still Identifiers starting with an underscore followed by an uppercase letter (2nd item on the list) Is that an oversight or...

Pasting here some info about versions distributed with Ubuntu LTS releases image | CMake | GCC | Clang --- | --- | --- | --- ~ubuntu:20.04~ | ~3.16.3~ | ~9.4.0~...

As far as **CUDA** is concerned, we are looking at a minimum of 12 or more recent. Release dates are documented at https://developer.nvidia.com/cuda-toolkit-archive We probably want to look at what...

Regarding **HIP** ROCm releases are tracked on https://rocm.docs.amd.com/en/latest/release/versions.html Docker images support is discussed at https://rocm.docs.amd.com/projects/install-on-linux/en/latest/how-to/docker.html#docker-images-in-the-rocm-ecosystem (includes link towards GH repo to build images and to DockerHub) We would want to...