Predict: The Next Generation Console Tech

[function]

And only a handful actually tailored them to those specs in making anything above and beyond hi-res xbox1 games.

Developers (well, everyone) knew 2 years in advance of shipping roughly what hardware was going into the 360. You are suggesting that developers will be able to essentially throw a switch and BAM xbox 3 game.

No third party would thank MS for secretly making their awesome launch games while expecting them to put unprepared titles into the market place against them. It's the kind of dick move that N64 era Nintendo would have made. Optimised multiplatform titles take time to develop and if Xbox 3 was going to be ready to ship in 10 months all third parties would have titles in full production now.

Unless MS has gained +10 in secrecy I just don't see this.

I don't get your point. You're saying pc/xb360 ports won't be enough to move hardware? If so, I agree that those by themselves will not be enough. There will have to be some sign of what is to come that DOES take advantage of the hardware in some measurable degree above and beyond xb360 to really get sales moving.

The point was that you can't just pull hi-tech PC games over to your new console if PC games have become low-tech (so to speak) console ports. Quake 4 was significantly beyond the Xbox. Rage* is perfectly at home on 360.

*I actually think Rage is very impressive.

As for it being a "stupid" idea or not, as long as MS is making money and taking steps to ensure they have a profitable future, I don't think they care if some would think of it as "stupid".

It's stupid because it'd be ensuring they fumble the transition and prematurely kill the cash cow that the 360 has turned in to.

Edit: Anyway, this is OT so we'd better leave it at that.

 

[homerdog]

Atleast we'll finally see tessellation give push...

This is actually a big deal. We will see massive improvement in geometric detail, and it will look startlingly good compared to what we're used to (see the Unigene Heaven demo).

I suspect console ->PC ports in the next gen to have DX11 as the minimum requirement because the renderers will be built from the ground up with tessellation and compute shaders in mind.

 

[TheChefO]

Hmmmn i dunno Chief... a 50% difference in performance could make more of a difference than we realise, especially given the console releases will be less than a year apart. It would certainly make more of a difference in multi-platform ports between the two consoles, and considering MS barely has a first party, i would expect it to be in their interests to ensure that they are within at most 10% of the performance of their competitor, so as to repeat the same situation they had with multiplat games this gen. If there's a significant different in muliplat games next gen, i think it could hurt MS more if it was in Sony's favour than the other way around.

But ultimately it'll come down to what the customer values more, sticking with the same software and online platfom, or the graphics/performance of the two consoles.

I'm sure ideally they would want to hold performance parity (or as close to it as possible) but there are other considerations such as launch time-frame and price.

The further we progress in graphics the more difficult it is to see advantages or disadvantages (without side by side comparisons breaking down every detail for those that care enough to see). Consider it took this long for Nintendo to start losing sales when they were using decade old technology!

Let's also not forget that Sony would have to then make a choice (if they were indeed late to launch) in choosing to go with the more powerful/bigger spec or to save those extra costs and match the existing console on the market.

Sony may just decide to pocket the difference and match performance to shave costs. It isn't as though Sony is swimming in cash these days ...

 

[TheChefO]

Developers (well, everyone) knew 2 years in advance of shipping roughly what hardware was going into the 360.

And just about a year ago word got out that xb720 kits were arriving at EA...

2 years later (~holiday 2012).


As for other pc games not pushing the envelope ... they have started to up the ante ... BF3 (and the Samaritan demo from Epic are two examples). The refresh (BF Bad Company 3 built on the new frostbite engine) would be a perfect game to showcase for a launch title in 2012.

I don't know about you, but I'm not expecting some radical departure in either Sony or MS. I expect both machines will be easy to code for (relatively speaking) and will allow for quick ports to flesh out the launch lib. A year or two after launch (as we saw this gen) and I suspect we will start to get titles that push the limits and grow from there.


Your using quake 4 as an example of something impossible on xbox and therefore a good showcase of what is possible on xb360 I think is a bit misleading. It was one title that didn't do anything especially impressive FWIR. The vast majority were simple ports of xbox & pc games that didn't show a clear advantage above and beyond xbox aside from resolution in rending and textures.

I expect that the vast majority of the launch lineup will be similar this go round for xb720.

They will likely be xbox360 games with better textures and higher res rendering (1080p). One or two may stand out, just as was the case last go round.

 

[Prophecy2k]

I'm sure ideally they would want to hold performance parity (or as close to it as possible) but there are other considerations such as launch time-frame and price.

The further we progress in graphics the more difficult it is to see advantages or disadvantages (without side by side comparisons breaking down every detail for those that care enough to see). Consider it took this long for Nintendo to start losing sales when they were using decade old technology!

Let's also not forget that Sony would have to then make a choice (if they were indeed late to launch) in choosing to go with the more powerful/bigger spec or to save those extra costs and match the existing console on the market.

Sony may just decide to pocket the difference and match performance to shave costs. It isn't as though Sony is swimming in cash these days ...

That's true but i wouldn't be surprised to see Sony spending big on their next console if it means securing and/or bolstering their position in the console space. Network products and services are one of Sony's most important divisions at the moment, more so i would say to Sony than xbox is to MS (considering MS' main meat and potatoes has and will always be windows). Sony isn't afraid to invest big when it's necessary, even post crazy Ken Sony. We've seen that with their recent Sony Ericsson buyout. MS would be foolish to underestimate that.

I don't think they will though. I see this coming gen becoming the most aggressive console war between platform holders yet (fanboys too ). With the rise of stuff like Apple's iOS and other forms of entertainment all vying for consumer's disposable income, there's no reason for any of the big three HW manufacturers not to be more aggressive than ever.

 

[wsippel]

That's true but i wouldn't be surprised to see Sony spending big on their next console if it means securing and/or bolstering their position in the console space. Network products and services are one of Sony's most important divisions at the moment, more so i would say to Sony than xbox is to MS (considering MS' main meat and potatoes has and will always be windows). Sony isn't afraid to invest big when it's necessary, even post crazy Ken Sony. We've seen that with their recent Sony Ericsson buyout. MS would be foolish to underestimate that.

I don't think they will though. I see this coming gen becoming the most aggressive console war between platform holders yet (fanboys too ). With the rise of stuff like Apple's iOS and other forms of entertainment all vying for consumer's disposable income, there's no reason for any of the big three HW manufacturers not to be more aggressive than ever.

Because that worked amazingly well the last two times.

 

[TheChefO]

That's true but i wouldn't be surprised to see Sony spending big on their next console if it means securing and/or bolstering their position in the console space. Network products and services are one of Sony's most important divisions at the moment, more so i would say to Sony than xbox is to MS (considering MS' main meat and potatoes has and will always be windows). Sony isn't afraid to invest big when it's necessary, even post crazy Ken Sony. We've seen that with their recent Sony Ericsson buyout. MS would be foolish to underestimate that.

I don't think they will though. I see this coming gen becoming the most aggressive console war between platform holders yet (fanboys too ). With the rise of stuff like Apple's iOS and other forms of entertainment all vying for consumer's disposable income, there's no reason for any of the big three HW manufacturers not to be more aggressive than ever.

I think Sony and MS will both be relatively conservative in their spec.

They will want to be break even as soon as possible (2 years max) and that will put a cap on die size and performance.

I'm not sold that Sony will want to eat the cost of a bigger die and I'm not sold that they want to be seen as the expensive AND late console again.

By MS making their move in 2012 (If that is indeed how things go) it will force Sony's hand and they will likely want to have performance parity with xb720 while keeping costs as low as possible.

I'll be looking for one or two big exclusive deals out of Sony and MS to make their next consoles onto their customers buy list. These exclusives would be relatively cheap to sway customers vs extra silicone over the life of the console and the ROI should prove better with the right exclusive(s).

 

[DuckThor Evil]

New 28nm GPUs are expected by 1Q 2012. There is plenty of time to improve yields until the end of 2012. 32nm are already in mass production at GF, and IBM new Power7+ is suppose to be out by mid-2012 @32nm.

These are good points. I don't think launching 2013 vs 2012 would give the manufacturers any significant benefits on what you can put into the boxes. Just like it didn't give any benefits in 2006 vs 2005 as far as the chips were concerned.

We have to also remember that it's not only about the launch year, but the entire lifespan of the consoles, the process nodes will most likely be the same for two manufactures throughout the lifespan whether the span in 2012-2020 or 2013-2020. It should probably be 2012 vs 2014 starting point, if you want some major differences and 2014 is a lot later than 2012. The more I think about it the more 2012 starts to feel like the most optimal launch window.

 

[Vice]

What's that birdie? You predict Steve Ballmer will be the opening act (night before) for the CES event in January. Something about the future of Cloud (content interaction), the pending Windows 8 launch timeframe and a new focused scope towards mobile devices. Maybe demoing these "new possibilities" between the mobile device and a nice charade kit. Just saying...

^^^ Sorry for the large text.

 

[DuckThor Evil]

If you'r referring to me, I'm not particularly predicting anything or even speculating at what event they might announce it, just saying that 2012 kinda makes sense.

 

[fehu]

They will want to be break even as soon as possible (2 years max) and that will put a cap on die size and performance.

If i rember right the 360 was even in the first 2 years (or 3?), but this time with so many live accounts, and the live itself making money selling services and movie/games i see MS having more budget for big die and memory.

 

[AlNom]

Anyone knows if there is an expected memory density increase from what's availiable in 2012 vs 2013?

The plans for GDDR5 haven't been updated for public consumption in a long time.

Xbox360 used 90nm, which became available in retail GPU a month after its launch.

Which kinda ended up being very expensive due to shit yields and a GPU with massive heat issues. A repeat of this would be... asinine to say the least.

New 28nm GPUs are expected by 1Q 2012.

Said GPUs have A) binning possibilities, and B) aren't produced or stocked up for a 1 million unit launch.

32nm are already in mass production at GF, and IBM new Power7+ is suppose to be out by mid-2012 @32nm.

Power7 is a monolithic design, and I'm not so sure this is even a high volume one at that (compared to console), so I don't know what the relevance is here. It's hardly inexpensive.

Regarding GF, well.... just FYI:
http://www.anandtech.com/show/4894/amd-confirms-32nm-yield-issues-at-global-foundries

Not the most reassuring.

There is plenty of time to improve yields until the end of 2012.

Just a few comments for anyone's imagination/information:

The process can clearly improve, but new nodes lag behind in yield compared to the previous node for quite awhile and by a fair margin, especially if you do want a high performance design with high transistor counts, which will be that much more sensitive to defects in the production. There are physical defects that don't actually shrink by changing node because the issue is with the wafer...

So you may get more chips per wafer, but how many of those are going to be functional compared to the previous node, which has had a far longer time to mature?

And also consider the fabrication capacity, which is almost certainly much lower than the previous node. Equipment is bloody expensive, and they don't just get rid of that when the new node hits. They continue fabbing things on the old node if the machines aren't being used, that means the machines are wasted. So it's going to be offered at better rates. More product tape-outs on 28nm? Translation: Lower fabrication capacity per single design. That does not make your chip orders cheap.

The fact that TSMC has been taking so bloody long with 28nm, should make you pause with PR statements - the usual PR garbage. So the yield targets have surpassed 40nm for the same time period, that doesn't mean anything useful because we obviously have much better 40nm production compared to whatever time they are referring to in the above PR. There's nothing false if the 28nm yield now is only slightly better per comparative time frame. There's nothing saying that it is significantly better.

2013 is a far more viable timeframe for mass manufacturing a part that has no possibility of binning and still has decent clocks without breaking some desired thermal target.

 

[DuckThor Evil]

Launching with 28nm gpu in Q4 2012 would still be much more conservative than what they did with the 360 and the headstart they got with 360 was one of the best decisions they've ever made despite the early difficulties. With better cooling system those issues could have been avoided. The chips were fine, although in limited quantities for a while.

Yeah the yields wont be as good and it's going to be more expensive per unit etc., but the payoff could be huge and higher early costs would even out in the long term.

 

[Megadrive1988]

These are good points. I don't think launching 2013 vs 2012 would give the manufacturers any significant benefits on what you can put into the boxes. Just like it didn't give any benefits in 2006 vs 2005 as far as the chips were concerned.

We have to also remember that it's not only about the launch year, but the entire lifespan of the consoles, the process nodes will most likely be the same for two manufactures throughout the lifespan whether the span in 2012-2020 or 2013-2020. It should probably be 2012 vs 2014 starting point, if you want some major differences and 2014 is a lot later than 2012. The more I think about it the more 2012 starts to feel like the most optimal launch window.

I tend to agree.

 

[AlNom]

Launching with 28nm gpu in Q4 2012 would still be much more conservative than what they did with the 360 and the headstart they got with 360 was one of the best decisions they've ever made despite the early difficulties.

Would it? TSMC started volume production of 90nm at the tail end of 2003. How is this comparable at all to 28nm in terms of absolute time? In terms of complexity, 28nm is a worlds difference of issues and problems compared to 90nm.

 

[Proelite]

Would it? TSMC started volume production of 90nm at the tail end of 2003. How is this comparable at all to 28nm in terms of absolute time? In terms of complexity, 28nm is a worlds difference of issues and problems compared to 90nm.

Why did Ati and Nvidia take nearly two years to move on to 90nm? Just curious.

You're saying that even more than half a year after 28nm GPUs are out next year, the process might not be mature enough for console use?

Bummer.

 

[Proelite]

You've got no chance of 6990 or GTX 580.

Power, heat, cost....

7850/7870 at most imho...

Atleast we'll finally see tessellation give push...

If they're targeting 7870 in raw power then they'll need 6990s in the PC dev kits to simulate it because 7870 is slightly more powerful than 6970, and can do much more in a console.

I don't believe the rumors of the dual GPU, but if they are true then my guess is that they have a low clocked, gimped 6990 that's somewhere between 6970 and 6990.

 

[DuckThor Evil]

Would it? TSMC started volume production of 90nm at the tail end of 2003. How is this comparable at all to 28nm in terms of absolute time? In terms of complexity, 28nm is a worlds difference of issues and problems compared to 90nm.

Well I'm mostly going by the expected release date of 28nm GPUs from AMD and nVidia (which should be Q1-Q2 2012 if I'm not mistaken) in relation to the 90nm gpus back in the day.

The GPU in "720" should be smaller than AMDs "big dog" chip, so probably slightly less issues and there is an incentive for TSMC maybe to offer a decent deal for a contract that has them making tens of millions chips in the following years even if they have to bite the lip a bit in the beginning, or at least it should be a ground for a good compromise

I agree that 2013 in terms of manufacturing would be more optimal, but perhaps squeezing it out in 2012 would offer other benefits that would make it a better overall strategy? I'm still not betting money on it

 

[AlNom]

Why did Ati and Nvidia take nearly two years to move on to 90nm? Just curious.

You're saying that even more than half a year after 28nm GPUs are out next year, the process might not be mature enough for console use?

Bummer.

Timing of their products is one factor. If they already have a design for a node, you don't just shift it to the new node and call it a day. 110nm was a half-node from 130nm, so moving to 90nm would require a much more extensive redesign. e.g. X800 was 110nm, R5xx was 90nm. If you take G70 -> G71, you'll note that G71 was heavily modified (i.e. 30M transistors less) for 90nm.

Also 110nm was cheaper and mature - it's just an optical shrink of 130nm rather than a whole new process. There's a lot of risk involved when moving to a whole new node, and again, consoles are in a worse position because they don't have binning possibilities.

----

Now of course, folks were expecting 28nm for a long while now, but the continued delays are just bad for everyone. One question is whether Nintendo feels up to spending a ton of money here to push things out on the new node and risk a shortage of products.

Another is if they were even expecting to do so, especially given that TSMC has been delaying and delaying. 2012 is quite possibly the worst year to release anything (with the hope of pushing tech) because we are on the cusp of shifting nodes, and that doesn't get cheap or mature quickly. It's expensive with crap yields compared to the previous node.

Is it so shocking that IBM announced 45nm for their WiiU design back in June? They may have 32nm, but it's not exactly primetime for anyone but Intel.

 

[almighty]

If they're targeting 7870 in raw power then they'll need 6990s in the PC dev kits to simulate it because 7870 is slightly more powerful than 6970, and can do much more in a console.

I don't believe the rumors of the dual GPU, but if they are true then my guess is that they have a low clocked, gimped 6990 that's somewhere between 6970 and 6990.

7870 is dead on 6970 performance, They won't need 6990's in the dev kits because they the dev kits won't have the API Windows crap stopping them.