GPU acceleration: Squeezing every last drop of compute | VividQ