compile linux to PTX / nvidia ISA and run on a single GPU core sequentially or parallelize the code and run on multiple; check feasibility