Can't use `rand!` or `randn!` with a collection of elements on a `CuArray`

[kshyatt]

MWE:

julia> using CUDA

julia> A = CUDA.zeros(128);

julia> CUDA.rand!(A, 1:64)
ERROR: Scalar indexing is disallowed.
Invocation of setindex! resulted in scalar indexing of a GPU array.
This is typically caused by calling an iterating implementation of a method.
Such implementations *do not* execute on the GPU, but very slowly on the CPU,
and therefore should be avoided.

If you want to allow scalar iteration, use `allowscalar` or `@allowscalar`
to enable scalar iteration globally or for the operations in question.
Stacktrace:
 [1] error(s::String)
   @ Base ./error.jl:35
 [2] errorscalar(op::String)
   @ GPUArraysCore ~/.julia/packages/GPUArraysCore/aNaXo/src/GPUArraysCore.jl:151
 [3] _assertscalar(op::String, behavior::GPUArraysCore.ScalarIndexing)
   @ GPUArraysCore ~/.julia/packages/GPUArraysCore/aNaXo/src/GPUArraysCore.jl:124
 [4] assertscalar(op::String)
   @ GPUArraysCore ~/.julia/packages/GPUArraysCore/aNaXo/src/GPUArraysCore.jl:112
 [5] setindex!
   @ ~/.julia/dev/GPUArrays/src/host/indexing.jl:58 [inlined]
 [6] rand!(rng::Random.TaskLocalRNG, A::CuArray{Float32, 1, CUDA.DeviceMemory}, sp::Random.SamplerRangeNDL{UInt64, Int64})
   @ Random ~/.julia/juliaup/julia-1.11.7+0.x64.linux.gnu/share/julia/stdlib/v1.11/Random/src/Random.jl:273
 [7] rand!
   @ ~/.julia/juliaup/julia-1.11.7+0.x64.linux.gnu/share/julia/stdlib/v1.11/Random/src/Random.jl:268 [inlined]
 [8] rand!(A::CuArray{Float32, 1, CUDA.DeviceMemory}, X::UnitRange{Int64})
   @ Random ~/.julia/juliaup/julia-1.11.7+0.x64.linux.gnu/share/julia/stdlib/v1.11/Random/src/Random.jl:265
 [9] top-level scope
   @ REPL[2]:1

[giordano]

What do you mean "on a subset of a CuArray"? 1:64 is the collection of elements to draw the random elements from, not the range of indices of A: https://docs.julialang.org/en/v1/stdlib/Random/#Random.rand!

Populate the array A with random values. If S is specified (S can be a type or a collection, cf. rand for details)

As an alternative to rand! you can do

julia> using CUDA, Random

julia> A = CUDA.zeros(8);

julia> A .= rand.((1:4,))
8-element CuArray{Float32, 1, CUDA.DeviceMemory}:
 4.0
 4.0
 3.0
 1.0
 3.0
 1.0
 4.0
 4.0

and if you really want to slice the array:

julia> A = CUDA.zeros(8);

julia> @views A[1:4] .= rand.((1:4,))
4-element CuArray{Float32, 1, CUDA.DeviceMemory}:
 2.0
 3.0
 3.0
 4.0

julia> A
8-element CuArray{Float32, 1, CUDA.DeviceMemory}:
 2.0
 3.0
 3.0
 4.0
 0.0
 0.0
 0.0
 0.0

[kshyatt]

Sorry, 🤧 brain struck again. You're right and I'll fix the title, it would be nice to support this natively.