Optimising GPGPU Execution Through Runtime Micro-Architecture Parameter Analysis (Full Report)