Why would I make that promise when I have never claimed anything like that or anything close to it. All I have argued for in this thread is that the 1060 3GB does just fine against the RX 470, and whilst that may change in the future, there is currently no data to clearly indicate that it will do so. In other words I haven't made a single prediction, quite the contrary I have warned others about making predictions based on feelings instead of data.
You on the other hand have been constantly arguing with conviction that 3GB will without a doubt become a bottleneck (example
1,
2,
3,
4,
5).
So the real question is are you willing to make the following promise right now:
"The 1060 3GB will be significantly VRAM limited vs. the RX 470 4GB in 2 to 3 years time. Significantly meaning that it loses by a significant margin (more than 10%) in the majority of games, due to a lack of VRAM (and not due to other issues such as DX12/Vulkan)."
If you make that promise, then I will gladly make a promise that actually matches my claims in this thread, which would be the following:
"The 1060 3GB may or may not be significantly VRAM limited vs. the RX 470 4GB in 2 to 3 years time, although there is currently no data to clearly indicate it one way or the other"
Then we can both promise to eat cat food if our promises turns out false as is tradition on these forums (hi Moonbogg).
Also I think it's hilarious that when I ask you for 460 768MB vs 1GB you actually think that 8800 GT 256MB vs 9600 GT 512MB is a suitable substitute. The whole point of looking at the 460 numbers is due to the fact that just like the 1060 3GB vs the RX 470 4GB we are looking at a comparable ratio of VRAM. Just admit it you made a claim that you can't back up and now you are trying your hardest to deflect.
But hey let me do your job for you,
here is the original 460 review from TPU, and
here is that latest review that included the 460 from TPU. The 768MB version started out being 8% slower at 1200P and ended up being 11% slower at 1200P, or in other words a loss in performance of 3%. Hardly what I would call collapsing. And mind you that over 80% of the games in the benchmark suite were replaced with newer ones, so the lack of performance loss wasn't due to not testing newer games.
Regarding the Guru3D review, I said that there wasn't much of interest to report because their number matched up with what we had already seen at that point, in other words nothing new.
And this may sound weird to you, but I actually look at the numbers in reviews not just the final conclusion. Reason being that you will often find idiotic statement in conclusion such as this one: "again I would advise 6GB as I feel 4GB+ is the norm for proper mainstream gaming anno 2016", which is obviously incorrect given that only one of the games out of 8 that Guru3D tested actually showed any kind of VRAM limitation (Hitman). If only 12.5% of the games you're testing this year shows something then it's obviously not the norm for said year.
And no their conclusion is not simply to buy the 6GB version, their conclusion is to buy the 6GB
if you are playing at 1440P (something I agree with) or
if you have the extra cash, the latter obviously being a shitty argument, since you can use that line of reasoning to argue that people should get the Titan X.