The idea is to couple the central processing units (CPUs) cranked out by Intel and AMD, which can movenimbly from one kind of task to another, with the GPUs put out by AMD and Nvidia, which are good at grinding through vast numbers of tasks at the same time.