Debugging float SIMD Intrinsics via llvmcall

It needs to be Tuple{NTuple{4, VecElement{Float32}},} for the input argument type. Things are a bit inconsistent when you can use a tuple and when you need a tuple type.

1 Like