MethodError: no method matching gemm!, It looks like |>gpu cannot manage arrays resulting from view and reshape

Specific Domains GPU
1

My julia version: 1.9 beta4

It looks like |>gpu cannot manage arrays resulting from view and reshape. cu() can it!

When I run the code down, I get this error:
Exception has occurred: CompositeException
TaskFailedException

nested task error: TaskFailedException

    nested task error: MethodError: no method matching gemm!(::Val{false}, ::Val{false}, ::Int64, ::Int64, ::Int64, ::Float32, ::CuPtr{Float32}, ::CuPtr{Float32}, ::Float32, ::CuPtr{Float32})
    
    Closest candidates are:
      gemm!(::Val, ::Val, ::Int64, ::Int64, ::Int64, ::Float32, !Matched::Ptr{Float32}, !Matched::Ptr{Float32}, ::Float32, !Matched::Ptr{Float32})

etc.

Solution:
replace: final_arr_gpu = (final_arr)|>gpu
with: final_arr_gpu = cu(final_arr)
Alternative solution:
before to send final_arr to gpu, add line: final_arr = collect(final_arr)

Code running |> in error:
using Flux
using CUDA

function testgpu()

Set the parameters

Generate Array with view and reshape

arr = rand(2, 3, 4)
sub_arr = view(arr, :, 1:2, :slight_smile:
final_arr = reshape(sub_arr, size(sub_arr)…, 1)
#final_arr = collect(final_arr)

@show typeof(final_arr) # Array{Float32, 4}

Define the model

model = Chain(
Conv((2, 2), 4 => 16, relu),
Flux.flatten,
Dense(16, 32, relu),
Dense(32, 1)
) |> gpu

Move the data to the GPU

final_arr_gpu = (final_arr)|>gpu

Run the model on the GPU

output = model(final_arr_gpu)