Why is this? If temps and power are good why do the boost clocks get throttled more on a ref cooler?
Edit: to that point, my reference cooler evga 980 ti will hit consistent 1415 boost clocks, and a stable 8000 on the memory. So I would imagine a titan x doing that would easily beat a 980 ti doing 1450 boost clocks and less than 8000 on the memory.
In the case of Titan X vs 980Ti, the Titan has a fully enabled GM200 chip which means there's a little bit more there that needs power delivered, and more of the die space thusly generating heat. Since the 980Ti is cut down slightly it can use more of that same allotted power to push for higher stable overclocks. And with marginally less heat produced on the core of the chip the mostly adequete reference cooler goes that extra mile on the 980Ti.
Closing the gap on a Titan X is a matter of hitting those higher boost clocks consistently at a rate that sufficiently trumps the Titans ability. Based on those above factors reference vs reference, that gap is closed from say a native 8-10% gap to something more like 2-4%.
But adequate as the reference cooler is, it still runs into that 84 degree threshold that adversely affects some of the real boosting potential. You'll see the cards with the best aftermarket coolers for instance hitting not only a higher boost clock than reference, but hitting that clock and mostly flat-lining there, without that wider frequent up and down range you would get with a lesser cooler.
Nvidia does themselves a great disservice by not allowing their partners to put aftermarket coolers on the Titan. But it's just as well because we get to feel better about our purchase where a reference 980Ti is within spitting distance of a Titan X out of the box, and non-reference models blow by it, all for less than $700.
