I worry if they don't wait for 10nm, 14nm isnt enough to offer a next gen experience. It's maybe 2X performance per watt over 28nm. Perhaps they aim for a 200W console that they can shrink quickly with 12-18 months, otherwise we're only going to get a 6-8 Teraflop console.
Predict: Next gen console tech (9th iteration and 10th iteration edition) [2014 - 2017]
- Thread starter Shifty Geezer
- Start date
- Status
Only?? 8 TF is what I'd consider very optimistic right now. Then again we really don't even know when these new babies will come out yet.
Michellstar
Regular
if it's forward compatible and you get a refresh each new node, why wait?? to 10nm
Basically, a AMD Fury (8Tflops around 200W currently on 28nm).
If that's the case, sure I have no problem with that. I'd actually prefer that.
Fury Nano is 8.2 Tflops with max tdp of 175W...
upnorthsox
Veteran
and 600mm2!!!
On 28nm, it'll be less than 200mm² on 14nm. Plenty of room for more CUs and 8 Zen cores.
Cheers
upnorthsox
Veteran
How do you figure that?
9th gen console with 8-16GB of memory? Urgh. And those 8 Zen cores iwll take up A LOT of those 350mm2.
Not realistic IMO.
The apu's on current consoles are around 350mm^2.
Not closer to 300mm^2?
Yes, 14nm is basically 20nm density-wise due. There's a little gain, but not significant. 14nm would allow a higher clock though. So we may see fewer shaders, but at a higher clock (maybe 1.2-1.3 GHz).
I would only expect two stacks of HBM for a console so a max of 16GB (Two HBM stacks) so there would be some die-area savings there as well by cutting the interface size.
The big question, imo, is would anyone go back to a split memory. 1 or 2 stacks of HBM and large DDR4 pool. If we're going with only HBM, I can't see the consoles going higher than 16, mayyybe 32. But with a split memory pool, I think we could see something like 16/32 GB DDR4 with 8/16 GB of HBM.
Even though I would like a console in the 10+ Tflops range around 2018 I think it is more realistic we are going to get something in the 7-8.5 Tflop range. Probably something like 24GB of unified HBM.
I just don't see either of these console makers wanting to veer a whole lot on energy consumption or a $400 price point.
It's possible that both drag this generation out to 2019 or 2020 and we end up with more powerful systems but I don't think Microsoft will want to wait that long to rain on Sony's PS4 parade.
I just don't see either of these console makers wanting to veer a whole lot on energy consumption or a $400 price point.
It's possible that both drag this generation out to 2019 or 2020 and we end up with more powerful systems but I don't think Microsoft will want to wait that long to rain on Sony's PS4 parade.
I think it would be a mistake to still have mechanical drives for the next generation of consoles. Not to mention the performance imbalance within in the actual hardware. I'd rather them compromise on processor costs to ensure flash storage.
I think in a few years it will be feasible to have an initial offering of a 500GB flash storage console. Probably also a 750 GB option for extra $100.
Why would MS or Sony want to release a new console so soon? With the slow down and diminishing returns on new hardware, how are they going to sell new consoles and make people want to upgrade to them? Even a 8TF console won't be a massive enough upgrade for a new gen and it would cost a fortune to make at 14nm compared to the current ones, especially since they would need to support both consoles for a couple of years. That is on top of the longer development time for games, changing hardware often doesn't seem economical any more. Also graphics have reached a point where its really hard to be WOW'ed by the new gen, its just going to be more AA, better texture and with more realist lighting than last gen but nothing like ps1 to ps2 to ps3.
To get a decent jump, a gen worth of jump would mean 8x the processing power and 8x the memory. When 64GB of ram is economical and a >12TF graphics card is <$200. then we'd actually get performance worth a new gen in a console. Even then, I don't know if they would be able to market graphics good enough to justify a new gen.
To get a decent jump, a gen worth of jump would mean 8x the processing power and 8x the memory. When 64GB of ram is economical and a >12TF graphics card is <$200. then we'd actually get performance worth a new gen in a console. Even then, I don't know if they would be able to market graphics good enough to justify a new gen.
The industry is in a bit of a bind with the console makers launching underpowered hardware after such a long previous generation. Xbox One will probably run the latest third party AAA games for another year before graphical compromises outside of just fps and resolution become apparent. What then? Will Microsoft be willing to continue with Xbox One until 2019 or 2020 to be able to make massive leap spec wise?
I guess the big benefit this gen is that consoles have become less expensive faster and people are buying them sooner which is good for game developers.
- Status