Hardware for data crunching and large arrays

[Woolley89]

I’m looking to up grade my hardware specifically for running autohotkey scripts. These scripts have to calculate many averages and other functions on every element of every array, but its taking ages to crunch the results. There’s a 3 dimensional cube testing combinations and each of those combinations have 40-70 arrays with 2 years of data accounting for each minute (2*365*24*60 = 1,051,200 elements in each* array in each combination)

The combinations dimensions are 50,128,12
Memory is not much of a problem I don’t think since each array is cleared after each combination is calculated in the grid. Keep in mind there are many functions being executed on every element of each array.

My question is what factors should I consider when buying a cpu or motherboard so I can run these simulations quicker? Would a duel cpu motherboard be a good option? Were crunching this data on a blockchain network, but we want to make this as centralized as possible.

[mcl]

You should consider learning C. Entire PC built out of Threadrippers running AHK can't compete with C program running on potato.
As an alternative to C you may try LuaJIT or Cython.

[swagfag]

but if u do want to throw hardware at ahk in particular, get the one with the highest sustained single core clock count
this will net u virtually no gains whatsoever in comparison with improving ur number crunching algorithms or simply rewriting the logic in native code. u should probably just do the latter and only then perhaps write an additional wrapper for use with ahk, if that even remains a requirement at that point(u can use OpenMP / AMP or any other highly optimized cpp math lib out there)