imo Vega 11 is the smaller chip and is a refresh of Polaris 10 with Vega GFX9 architecture . This architecture is supposed to bring a new Compute unit design and improvements to efficiency and throughput. AMD seems to be calling it NCU (New Compute unit or Next Compute unit). The Vega NCU combined with 20% higher clocks as seen in Vega 10 and probably G5X should bring a significant improvement in performance for Vega 11 over Polaris 10. imo AMD will be targetting the USD 199 - USD 299 price segment for Vega 11 and its variants (partially cut Vega 11 ). If the perf/sp (IPC improvements) are 15-20% this chip could turn out to be very impressive.
Vega 10 is the big flagship chip and the first GPU with Vega GFX 9 IP . Vega 10 will launch in Q1 2017. Vega 11 is most likely to come in late Q2 2017 (towards mid 2017).
Interesting, and definitely a possibility. Somewhat un-AMD not to rebadge endlessly though, see Pitcairn the Immortal 2012-2016, and Hawaii 2013-2016. Tonga replacing Tahiti happened, however, so what you envision is very possible. Hopefully this time around it will be less underwhelming (having MORE and not less bandwidth this time around would be a step in the right direction).
Overall disappointing though, as I struggle to imagine a 4096SP Vega competing with Titan XP / 1080 Ti. My optimistic approach is that it can be in-between 1080 and Titan XP if engineered and driven right, but I'm not even expecting that, just hoping as a best case scenario.
Is AMD abandoning the large die size? As we know, Polaris 10 is similar to Tonga wattage. AMD had 2 chips above Tonga is die size / performance / wattage: Hawaii and Fiji. For your theory to be correct, AMD is no longer pursuing a Fiji equivalent. Vega 10 would be the Hawaii of the 14nm generation, Vega 11 would be Tonga, and we will have to wait for Navi for a GP102/100 (GV104?) competitor.