fail to generate large CUDA Array

[ArrogantGao]

I found that I can generate large CuArray, it seems that for generate value type, size of the array is limited below an Int32 number.

Here is a MWE.

using TropicalGEMM
using CUDA

CUDA.zeros(TropicalF32, 2^16, 2^16)

Error message:

ERROR: InexactError: trunc(Int32, 4294967296)
Stacktrace:
  [1] throw_inexacterror(::Symbol, ::Vararg{Any})
    @ Core ./boot.jl:750
  [2] checked_trunc_sint
    @ ./boot.jl:764 [inlined]
  [3] toInt32
    @ ./boot.jl:801 [inlined]
  [4] Int32
    @ ./boot.jl:891 [inlined]
  [5] convert
    @ ./number.jl:7 [inlined]
  [6] cconvert
    @ ./essentials.jl:687 [inlined]
  [7] macro expansion
    @ ~/.julia/packages/CUDA/1kIOw/lib/utils/call.jl:222 [inlined]
  [8] macro expansion
    @ ~/.julia/packages/CUDA/1kIOw/lib/cudadrv/libcuda.jl:5139 [inlined]
  [9] #735
    @ ~/.julia/packages/CUDA/1kIOw/lib/utils/call.jl:35 [inlined]
 [10] check
    @ ~/.julia/packages/CUDA/1kIOw/lib/cudadrv/libcuda.jl:35 [inlined]
 [11] cuOccupancyMaxPotentialBlockSize
    @ ~/.julia/packages/CUDA/1kIOw/lib/utils/call.jl:34 [inlined]
 [12] launch_configuration(fun::CuFunction; shmem::Int64, max_threads::Int64)
    @ CUDA ~/.julia/packages/CUDA/1kIOw/lib/cudadrv/occupancy.jl:61
 [13] launch_configuration
    @ ~/.julia/packages/CUDA/1kIOw/lib/cudadrv/occupancy.jl:56 [inlined]
 [14] (::KernelAbstractions.Kernel{…})(::CuArray{…}, ::Vararg{…}; ndrange::Tuple{…}, workgroupsize::Nothing)
    @ CUDA.CUDAKernels ~/.julia/packages/CUDA/1kIOw/src/CUDAKernels.jl:107
 [15] fill!(A::CuArray{Tropical{Float32}, 2, CUDA.DeviceMemory}, x::Tropical{Float32})
    @ GPUArrays ~/.julia/packages/GPUArrays/uiVyU/src/host/construction.jl:22
 [16] zeros(::Type, ::Int64, ::Vararg{Int64})
    @ CUDA ~/.julia/packages/CUDA/1kIOw/src/array.jl:773
 [17] top-level scope
    @ REPL[14]:1
Some type information was truncated. Use `show(err)` to see complete types.

notice that 4294967296 is exactly $2^{32}$.

[GiggleLiu]

Maybe, this should be an issue for CUDA.jl?

[ArrogantGao]

Maybe, this should be an issue for CUDA.jl?

For other data type such as Float32 it is fine, I am not sure is that a problem about generic data type.

Also does not works for Float64.

[ArrogantGao]

I modified the fill! function to make the zeros function works for TropicalF32 as follows

function Base.fill!(A::DenseCuArray{TropicalMaxPlus{T}}, x) where T <: CUDA.MemsetCompatTypes
         U = CUDA.memsettype(T)
         y = CUDA.reinterpret(U, convert(T, x.n))
         CUDA.context!(CUDA.context(A)) do
           GC.@preserve A CUDA.memset(convert(CuPtr{U}, pointer(A)), y, length(A))
         end
     A
end

However, permutedims! fail due to the same reason.

[ArrogantGao]

This problem has already been solve in master brach of CUDA.jl, JuliaGPU/CUDA.jl#2666.

[GiggleLiu]

Cool!