InvalidIRError when ndims(count) or ndims(prob) is >= 3 units smaller than ndims(A) #3
Description
Minimal example:
using CUDA, BinomialGPU
A = CUDA.zeros(Int, 2, 2, 2, 2)
ns = CUDA.fill(128, (2, 2))
ps = CUDA.rand(2)
rand_binomial!(A, count = ns, prob = ps)