GPU support

[parfenyev]

Hello,

I am trying to construct grid on GPU() according to documentation:

using FourierFlows

dev = GPU()
n, L = 16, 2.0
grid = OneDGrid(dev, n, L)

and I get the following error:

MethodError: no method matching OneDGrid(::GPU, ::Int64, ::Float64)
Closest candidates are:
  OneDGrid(::CPU, ::Any...; kwargs...) at ~/.julia/packages/FourierFlows/tVoz9/src/domains.jl:349
  OneDGrid(::Any, ::Any; x0, nthreads, effort, T, aliased_fraction, ArrayType) at ~/.julia/packages/FourierFlows/tVoz9/src/domains.jl:59
  OneDGrid(::Int64, ::Int64, ::Int64, ::T, ::T, ::Tx, ::Tk, ::Tk, ::Tk, ::Tk, ::Tfft, ::Trfft, ::T, ::Talias, ::Talias) where {T<:AbstractFloat, Tk, Tx, Tfft, Trfft, Talias} at ~/.julia/packages/FourierFlows/tVoz9/src/domains.jl:17

Stacktrace:
 [1] top-level scope
   @ In[14]:5
 [2] eval
   @ ./boot.jl:373 [inlined]
 [3] include_string(mapexpr::typeof(REPL.softscope), mod::Module, code::String, filename::String)
   @ Base ./loading.jl:1196

Probably, the documentation should be fixed.

[parfenyev]

The code works when moving to CUDA 3.3.6 (from 3.9). See also this discussion.

[navidcy]

Thanks! Let me have a look!