| JEMC said: One good thing about the GPU market is that neither Nvidia nor AMD (especially AMD) have the luxury to not go for the max they can, because they have a lot to lose. That said, I think the problem lies more in the rest of the industry than in AMD/Nvidia. The 14 and 16nm manufacturing processes aren't ready yet to produce the big chips that are needed for Fury/Titan class cards, GDDR5X won't be here until summer (we may still see it in board partner cards, but not reference models) and HBM2 seems to be too expensive for the mainstream market if what Raja Koduri said in the PCPer interview. I still think that we may get cards that are ˜10% faster than the current flagships at a lot lower power consumption and price, which seems small but it's still more than what we get from every new gen of Intel CPUs |
.I'm surprised how we're still concerned about yields when the metal pitch for 16nm is the same from 20nm and it's similar for 14nm too ...
@Bold Maybe so but I'm still more impressed by Intel's performance than any other since they have the process node with the highest transistor density in the world and I like some of the new features they bring in too. Suffice to say I look forward to Intel bringing AVX-512 including new x86 extensions along with something cool like the 3D XPoint on mainstream CPUs ... 
Essay Pro