PC Perspective published an interesting piece that explores why NVIDIA's Pascal generation features both a GP100 and a GP102 GPU. The GP100 is used by the Tesla P100 datacenter card, this model has a 600mm² die area, features 3840 CUDA cores, HBM2 memory and an ideal ratio of 1:2:4 between FP64:FP32:FP16 performance.
The GP102 on the other hand is used for the new Titan X, this consumer-level card swaps the HBM2 for GDDR5X memory and features the same 3840 CUDA core count in a 471mm² package. One big difference though is that this model has much wore FP64 and FP16 performance.
Anyway, that aside, it puts NVIDIA in an interesting position. Splitting the xx0-class chip into xx0 and xx2 designs allows NVIDIA to lower the cost of their high-end gaming parts, although it cuts out hobbyists who buy a Titan for double-precision compute. More interestingly, it leaves around 150mm2 for AMD to sneak in a design that's FP32-centric, leaving them a potential performance crown.