You'd have to be borderline retarded to not compare OC vs OC now when cards are available and people want to see the results, even if they will change in the future. People want to see them compared now, so they do it now and maybe they'll do it again later.
These vendors usually ship cards with overclocking utilities. AMD has overclocking tools and they know very well how the voltage is controlled on the card they made.
I can't think of a single gpu in years that has launched with no voltage control. I also have a very hard time accepting this as just some oversight by AMD. I believe it was purposefully locked and none of the vendors were given permission to allow voltage control. When nvidia limited voltage control, all hell broke loose. Nvidis voltage control today is very limited. Msi only allows 80mv on my 980. It's not a total lockdown but it is very very limited.
Some people say maxwell overvolts itself. But that isn't really true. It works within a range which is more complex. Because the voltage is dynamic, ppl say and think it overvolts itself. Not really, it is just dynamic and working within its ranges.
My point is, nvidia allows very little overvolting on maxwell. Adding .88mv on my 980 doesn't help my overclock at all, it has a negative effect.
Nvidia boost is sophisticated and doesn't work at one set voltage. Different clocks and different loads cause different voltages.
I think it is pretty obvious, at least to me, why and how maxwell overclocks like it does. Nvidia just didn't push the clocks, they were conservative and kept well within the sweet spot for the design. There is a range where the power consumption and clock curve is optimal. similar to a car engine and its hp to rpm curve. I believe nvidia clocked their chips in the middle of the curve while AMD pushed GCN to the edge for max performance.
Look at fermi, as power hungry as they were, the gtx470, 480, and 460s clocked like mad.....they had a lot of room left. Back then the node yielded chips with much lower clock speeds but we were seeing 20% overclocks easy, on stock voltages. 20% is low balling it, bad too.
Back then, there was headroom. Maxwell has headroom, fury not so much.
I believe AMD is clocked on the edge of the ideal zone and this prevents higher overclocking. With voltage, you would get more MHz but the power consumption would disproportionately shoot up.
When the nano launches, we can see if it holds true. It should be clocked lower, but if it is able to overclock at a higher percentage over fury x, then there is no more debating this.
Facebook
Twitter