Oliver Schulz

However we do this I think we should really aim to establish a single type tree for computing devices across the ecosystem, one that reaches down to different GPU (TPU,...

> The main advantage is we work pretty hard to navigate type wrappers in the most robust way we can so that if you grab a pointer from something you...

> @Tokazama : ArrayInterfaceCore. We don't have anything that merges device types and we account for a difference in CPU types that are built on different memory objects like a...

I'd like to get this moving again - @Tokazama can `ArrayInterfaceCore.AbstractDevice` evolve/change a bit as part of this effort?

Thanks @Tokazama . Ok I'll take a look - otherwise we can just provide a conversion between the device API in this PR and `ArrayInterfaceCore.AbstractDevice`.

Also, this PR is not just about checking if an array is a GPU array, but about checking what kind of device(s) a whole data structure is located on. Plus...

I'm not sure I've understood all the implications here - are you looking for somethings like a `VectorOfVectors` wrapped around an [`UnsafeArray`](https://github.com/JuliaArrays/UnsafeArrays.jl)?

But why can't we use `ReinterpretArray` if we have to keep the reference to `A` around anyway?

> A .= ntoh.(A) (10x slower) I wonder why ReinterpretArray is performing that much worse, compared to our custom wrappers. Well, probably can't be helped right now. In that case...

> @oschulz can you confirm TypedTable of VectorOfVectors is SIMD-able at all? SIMD for `VectorOfVector`'s columns in a `TypedTables.Table` is not straightforward, because there are two loops involved. But it...