sparse(I, J, V) doesn't support unsorted inputs

[CarloLucibello]

julia> using CUDA, SparseArrays

julia> I = [1, 1, 2, 3, 3, 4, 5, 4, 6, 4, 5, 6, 6, 6]

julia> J = [4, 6, 4, 5, 6, 6, 6, 1, 1, 2, 3, 3, 4, 5]

julia> A = sparse(I, J, ones(length(I)), 6, 6)
6×6 SparseMatrixCSC{Float64, Int64} with 14 stored entries:
  ⋅    ⋅    ⋅   1.0   ⋅   1.0
  ⋅    ⋅    ⋅   1.0   ⋅    ⋅ 
  ⋅    ⋅    ⋅    ⋅   1.0  1.0
 1.0  1.0   ⋅    ⋅    ⋅   1.0
  ⋅    ⋅   1.0   ⋅    ⋅   1.0
 1.0   ⋅   1.0  1.0  1.0   ⋅ 

julia> Agpu= sparse(I |> cu, J |> cu, ones(length(I)) |> cu, 6, 6)
6×6 CUDA.CUSPARSE.CuSparseMatrixCSC{Float32, Int32} with 14 stored entries:
  ⋅    ⋅    ⋅   1.0   ⋅   1.0
  ⋅    ⋅    ⋅   1.0   ⋅    ⋅ 
  ⋅    ⋅    ⋅    ⋅   1.0  1.0
 1.0   ⋅    ⋅    ⋅    ⋅   1.0
  ⋅    ⋅    ⋅    ⋅    ⋅    ⋅ 
 1.0  1.0  1.0  1.0  1.0   ⋅ 

julia> Array(Agpu) == A
false

[CarloLucibello]

Sorting by row index seems to fix this specific example:

julia> p = sortperm(I);

julia> I2, J2 = I[p], J[p];

julia> sparse(I2 |> cu, J2 |> cu, ones(length(I2)) |> cu, 6, 6)
6×6 CuSparseMatrixCSC{Float32, Int32} with 14 stored entries:
  ⋅    ⋅    ⋅   1.0   ⋅   1.0
  ⋅    ⋅    ⋅   1.0   ⋅    ⋅ 
  ⋅    ⋅    ⋅    ⋅   1.0  1.0
 1.0  1.0   ⋅    ⋅    ⋅   1.0
  ⋅    ⋅   1.0   ⋅    ⋅   1.0
 1.0   ⋅   1.0  1.0  1.0   ⋅