Jago Strong-Wright

I've just realised that the `AppleAccelerate` library is actually for M1 CPU acceleration so isn't the CUDA FFT equivalent we need here.

Perhaps we could convert to this: https://github.com/DTolm/VkFFT which supports hardware-accelerated FFT on CUDA, Metal and lots of others. It looks like that library is more performant than `cuFFT` as well.

Yeah we could just do it for Metal. I was just thinking it might be just as much effort as using it for all.

> I think it would be more effort if we take into account the need to rebenchmark a lot of cases. I also think it adds some risk... Yeah true,...

I've now had a play trying to wrap VkFFT with https://github.com/JuliaInterop/Clang.jl/tree/master but it is proving difficult given my inexperience with C. Does anyone working on Oceananigans have experience doing that...

weird that all the tests fail, when I run e.g. `RectilinearGrid(CPU(), FT, size=(16, 16, 16), extent=(1, 1, 1))` locally it works fine

If I try to just output the whole of the fields in the lat lon example then it fails to be able to output the free surface displacement with the...

Here's an example: ``` using Oceananigans grid = RectilinearGrid(arch, size=(20, 20, 20), extent=(1, 1, 1)) a_field = Oceananigans.Fields.Field{Center, Center, Center}(grid; indices=(:, :, 1:1)) model = NonhydrostaticModel(grid = grid, auxiliary_fields=(;a_field)) simulation...

> Is this still an issue? I've just come across this again so I think so yes. Might try and fix soon

I think this is the same bug as https://github.com/CliMA/Oceananigans.jl/issues/2770