Looks like my 1080Ti is holding up well..

[TheRealDeal]

Fermi was the last time Nvidia got a die shrink wrong. I think they went to big to soon so they got delayed. The performance was half decent though. I don't think Nvidia will make that mistake again hence why they are waiting on 7nm maturing before they jump on. Helps also that they are not under pressure to get on the new node as well.

 

[Alex Jeal]

It would be a very few interesting years if Nvidia messed up 7nm, but... has Nvidia ever messed up a die shrink?

My own expectation is that if 7nm gets any less than 50% performance boost from RTX2000 series then that would a failure

But even if the architecture is capable, they'll make far more money from releasing with a ~20-30% boost (3080 = 2080 Ti), and a diabolical $1400 Ti, then unlocking the lower tiers with SUPERs 9-12 months later...

 

[aoaaron]

It would be a very few interesting years if Nvidia messed up 7nm, but... has Nvidia ever messed up a die shrink?

My own expectation is that if 7nm gets any less than 50% performance boost from RTX2000 series then that would a failure

lmao.

why would they do do that when they can release a 20-25% card and make the same money with no competition.

 

[chrcoluk]

Interesting

Looks like Turing optimisation is a thing - the 1080ti is clearly dropping off. Was 2080 FE competitor but now it’s down to 2070 Super competitor.

The 1080ti isnt getting slower, just as is normal optimisation work is focused on newest gen so rather that gains vs the launch period. The only surprise for me was learning that AMD have a competitor to the 1080ti which I didnt think they had. I guess I am out of touch on the AMD gpu's.

Also the gap between a 2070 super and 2080 is much closer than e.g. the 1070 and 1080. The 1080ti will still trade blows with a 2080, just less favourably than at 2000 launch.