So you think cinematic framerates are acceptable now? The longer this gen goes, the worse it will get.
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
vg247-PS4: new kits shipping now, AMD A10 used as base, final version next summer
- Thread starter Globox_82
- Start date
So you think cinematic framerates are acceptable now? The longer this gen goes, the worse it will get.
Looking at games like TLoU etc, I'd say, "shit" is a very strong word to use. Things like IQ, framerate and resolution are definitely better on PC but what's being ringed out of a hardware that's 7 years old is nothing short of amazing.
Of course, we'd all like to see another leap of similar magnitude soon but these things have their own pace and drawing comparison between a highly obsolete tech and current high end tech doesn't change the "when"s and "what"s.
Yep. TSMC and GloFo aren't the only fabs. Mind you, there's also 32nm. Who knows what MS is up to.
Naturally, there would need to be a fair bit of time to migrate designs between nodes at different fabs if that's what they feel is necessary to do.
Nope.
R&D in 2012 is hardly grounds for a reliable process node & volume manufacturing for the end of 2013, especially for a non-mobile design.
I see the article writer has mentioned TSVs, but the CTO has not been quoted as saying that. It read more like he was regurgitating some of the Sony presentation.
The presentation looks at possibilities, but also acknowledges a number of issues in using state of the art 3D fabrication ("Test, Cost ,Heat ,Reliability , Multiple-supplier").
There are no guarantees they'll use it.
Sure, 28/32nm would have been cost-prohibitive this year compared to what they've already got at 40/45nm.
Next year, 28/32nm will be in a much better state as the fabs will be reaching much higher wafer output. That doesn't mean Sony needs a 32nm PS3. 40/45nm will still be cheaper and it's not like they're facing yield and heat problems.
With respect to the early tape-out, the time between then and launch would have been used for testing their design and trying to increase yields, on a process that actually exists for mass manufacturing, while also not competing with every mobile customer for limited wafer output (20nm).
edit:
A much more significant cost reduction would certainly be going to a single merged die, and for all we know, that may not be as good a proposition at 28/32nm - it may still be a relatively large die and the cost savings might not be significant enough for their targets. There would have to be a significant re-engineering of the Cell<->RSX connection anyway, so skipping 32nm might make sense if the die size isn't that small (hitting the limits of their I/O requirements would be ideal). Anyways, they will want to milk as much as they can at the current higher price while their current 40/45nm chips are serving them well.
You're wasting your time my friend, I and others have told him we won't be seeing 20nm or that TSVs isn't a guarantee but he keeps going on about it.
yurinka
Member
I'd say that for them to continue selling profitable PS2, PSP, PS3 and Vita during some months more would be better for them than to -likely to be sold at loss- PS4.
Specially if they can bump these sales with Vita and PS3 pricecuts if they still are profitable after it.
REMEMBER CITADEL
Banned
No. Some games are struggling, but you always had some games struggling. Others are fine.
Totally agree with this. I had a meeting with my boss recently and the issue of yields across all the major FABs came up. Some of the yield results for TSMC and Global Foundries were shocking. On the 28nm node both are struggling to hit above 40% yield on current AMD, nVidia products. Even Intel is really struggling with yields something which usually never happens with them, around 75% yield on 15nm node.
Reckless Onion
Member
i've already made the jump to pc cause this gen is going on far to long imo.
I'll still consider buying a console but they will need to seriously impress me (60fps/1080p)
I'll still consider buying a console but they will need to seriously impress me (60fps/1080p)
Well I would think it would be better to launch the PS4 while these other profitable systems are still selling so they can help make up for the losses of each PS4 sold. The launch of the PS4 should not greatly effect the sales of the previous systems.
However what may be a concern for them is having the PS3 and PS4 so close in price possibly? I have no idea what's going on with Sony and the PS3, but they had to have slightly bumped the price for a reason. Maybe they wanted more of a profit margin per console sold? I not sure, but I think they would prefer to create a nice price gap between their last gen/low end hardware and their new high end console. This way there is minimal overlap for the target markets of both products.
Also like Napalm_Frank pointed out, waiting too long can really hurt them in the long run. If they can get kits and good tools out in the hands of developers fast, they raise the chance for multi-platform titles leading on their console, which is always a good thing.
Vol5
Member
And here's why I think a Sony Meeting in Q1 is inevitable. PS3 price drop + PS4 announcement.
I'm also thinking the days of selling your console at a loss....those days are over. Sony will maximise their margins with an intelligent design thats powerful enough to deliver a visual punch but is relatively cheap to mass-produce. There will most likely be meltdowns because the 720 is more powerful, but if Sony can please Devs with a straight-forward design that allows them access to the metal and create games more easily than on a competitors system, it's fair to say multi-plat titles will look the same as any other anyway. The Vita is a great handheld and is sold at profit (edit: will be after 3 years). That model will carry over for PS4, maybe even sooner than 3 years.
I would be REALLY surprised if Sony didn't sell the ps4 at a loss. All the rumors indicate they are willing to take a hit on hardware again. You're right that they'll likely reach profitability faster than with the ps3.
It'll be sold at a loss, but not a very big one. I'm thinking $50 per console at the absolute max, which would also allow for quality hardware and not much of a financial strain. It'll be nothing like the $200+ loss on the PS3.
As for reveals I'm not expecting the PS4 at the Sony event next month but I am expecting an announcement before E3, same with Microsoft. The announcement will just be a name and a few teaser trailers, maybe tech specs and the console design. We won't see games running in real time until E3.
As for reveals I'm not expecting the PS4 at the Sony event next month but I am expecting an announcement before E3, same with Microsoft. The announcement will just be a name and a few teaser trailers, maybe tech specs and the console design. We won't see games running in real time until E3.
Vol5
Member
Which Sony event is next month?
Which is why I mentioned that they may do some form of subscription or instalment pay option contract with interest to offset any potential losses that may incur from buyers who want to buy their console outright. It's a long run game but the next gen will be around for the next 6 years at least.
pieatorium
Member
All the most recent have been though, Vita, WiiU and 3ds post price cut
imo it's impossible for any manucfacturer not to sell a new machine at a loss at launch... what matters is how much and how long. PS3 was truly a disaster in both areas. Not only the loss taken by machine was massive, but it lasted until the launch of the slim. Completely out of control.
jeff_rigby
Banned
Yet TSMC has already taped out 50 third party designs @ 20nm for volume production in 2014 or earlier. If the assumption is that yields will be smaller at 20nm and leakage is now a larger part of the consideration then why go with bulk and 20nm, stay at 28nm and go SOI which is supposed to reduce the leakage allowing lower transistor voltages which allows 28nm SOI to approach 20nm bulk.
I will agree that a console does not have the need for 20nm to save battery life, 32nm on SOI even is, I think, practical. Considering the design was likely finished in 2011 the wait to manufacture may only be tied to the Stacked DDR4 20nm availability which is the last half of 2013. So no rush to fab Thebe.
My points for 20nm take into consideration that AMD has been planning since 2007 for HSA and waiting till 2014 for Forges to be able to support; 20nm, TSVs & 3D stacking and has developed a large portfolio of building block IP to be manufactured with different processes and node sizes to be 2.5D and 3D assembled. This includes schemes for testing and alliances with other forges to allow common standards.
The issues with yields at smaller node sizes have been known since 2008 and one way to deal with it is by splitting up large designs into smaller sections and pre-testing. A factor in yields is also trying to include different "processes" or doping multiple different transister designs on the same chip, RF and Power transistors for instance require different doping.
Creating a GNB wafer at 20nm (Combination NB, SB, GPU, audio and power optimization circuits) could be used with other parts @ 28nm. It's my understanding that the GNB northbridge is the part that needs 20nm efficiencies and TSVs @ 20nm to connect to transposer that would have the wide IO ram stacks. GNB 3D or 2.5D to CPU packages @ 28nm is possible.
If you imagine the GNB wafer as the key to any design that any 4 CPU packages be they bulldozer or Jaguar can be attached you can see this as the key to building block designs and third party IP using the GNB.
TSMC @20nm is an all in house Fab from design to 3D assembly and TSMC has developed the tools without having to negotiate standards with other manufacturers; this probably put them ahead of other forges who attempt to standardize with others.
One of my faults is underestimating the time it takes to implement new processes which I see the advantages in using. I have a hard time understanding releasing 6 months before volume 20nm and 3D stacking is mainstream as I see the advantages and have read of the planning by everyone since 2008 for 2014. I assume the delay in releasing the PS4 is related to waiting for new processes like 20nm where 3D stacking is starting and full HSA designs are released. I assume Full HSA starts with new memory controllers that use 3D stacked wide IO memory.
bigtroyjon
Member
Was sitting in Jury Duty today waiting for them to eventually tell us the trail wasn't ready and we weren't needed and was thinking about a variety of random subjects. Next gen speculation popped into my head and I had this thought.
What if Microsoft realizes that they can't(or don't want to spend the money trying) compete with Sony when it comes to overall volume of 1st party software and because of that, they are going to outdo them in hardware in order to be the clear cut place to play 3rd party games?
Whenever I hear rumors or speculation, I like to look at the reasons behind the idea to see if there is any reason it would happen. To me this is the most logical reason why Microsoft would spend the money to out do Sony.
The opposite end of this spectrum would be Nintendo, who doesn't really care about 3rd parties at all since they can sell their hardware with the most valuable portfolio of ip in gaming.
What if Microsoft realizes that they can't(or don't want to spend the money trying) compete with Sony when it comes to overall volume of 1st party software and because of that, they are going to outdo them in hardware in order to be the clear cut place to play 3rd party games?
Whenever I hear rumors or speculation, I like to look at the reasons behind the idea to see if there is any reason it would happen. To me this is the most logical reason why Microsoft would spend the money to out do Sony.
The opposite end of this spectrum would be Nintendo, who doesn't really care about 3rd parties at all since they can sell their hardware with the most valuable portfolio of ip in gaming.
jeff_rigby
Banned
Proelite said:
This is an important point. You start with specs for what you want, in this case we hear from Sony CTOs 10X PS3 and 300FPS. Then AMD looks on it's roadmap and gives Sony options along it's roadmap several years in the future. Low power highly efficient XTV standby needed so Jaguar CPUs and biggest bang for the buck HSA efficiencies with the GPU. So in 2010 they give Sony a PC that simulates the final hardware. Sony and probably some third party game engine developers look at it and give an OK. The design is pretty much fixed at that point.bkilian said:
From that point forward AMD is implementing the same hardware in the roadmap they gave Sony. This is not a simple process as in that roadmap is other IP like memory that has to be available and wide-wifi radio and several other features to make the vision possible in the 2010 leaked Xbox 720 powerpoint. See the point, the hardware choices had already been made in 2010 at the latest.
I asked that question before and I was told no, no installs haha
interesting speculation - MS have certainly not pushed as many first party games later on this gen. And perhaps that is one reason Sony are still pushing unique exclusives, to try to differentiate to encourage people to buy.
I'd be curious to know how the economics work out. Margins on first party games are pretty good - you get the publisher's margin plus your own platform fee. So big sellers are probably profitable enough that you'd continue with them.
2-4 x faster bluray drive, potentially 8x more ram? don't think installs will go away
Sqrt_minus_one
Member
Definitely agree with this.
From a technical perspective it makes more sense to wait for 20nm and I hope both MS and Sony do, but from a business perspective they will be advantages going for 28nm.
More rapid time to market and the more mature process meaning higher yeilds but larger dies may mean that the 20nm process' small die advantage may not show up in the cost savings column.
The 20nm process will command a higher price per wafer and given that the cost of each die may end up being roughly the same if not higher in the short term at 20nm. Given this the only advantage of 20nm would be perhaps a 10% bump in clocks/performance and reduced heat output. These 2 advantages will have to be weighed against a more rapid time to market by going 28nm, I am not sure 20nm will win that fight.
also you need to factor in longer term cost reductions. pushing for 20nm now might mean higher initial costs *and* an inability to shrink processes to cost reduce for a long time.
going with a more conservative approach like 28nm means better yields now, and the ability to switch to 20nm in a year or two when the process is more mature, and cost reduce the console.
going with a more conservative approach like 28nm means better yields now, and the ability to switch to 20nm in a year or two when the process is more mature, and cost reduce the console.
jeff_rigby
Banned
Your only missing forges will be staying with 20nm for 3-4 years and 3D stacking infrastructure is designed for the same chip/pad configurations so staying with the same node/wiring/pad size makes it easier and having the same standardized building blocks easier to build testing equipment for sub assemblies.
These two finds with GNB @ 20nm in a 2013 Samara APU should be discussed. It's also seen that BOTH GF and TSMC will be producing it for 2013 and for only notebooks?????? GNB can/should be the core of both next generation consoles!
http://www.indeed.com/r/Rami-Dornala/e0704aad508659b2 said:
Actually 20nm means easier and less expensive shrinks.
For the next few shrinks after 20nm the wiring stays at 20nm but the transisters change to smaller sizes.http://www.xbitlabs.com/news/other/display/20121204221042_Globalfoundries_to_Start_14nm_Tests_in_Q1_2013_Company.html said:
The whole use of a subscription model wouldn't work for the ps4 unless it was backwards compatible or if PlayStation plus and Xbox live were identical (ie pay for multiplayer and access to services). This statement is also assume Sony would copy ms subscription model.
systemfehler
Member
I do hope for 200+ where the APU takes ~160W (120 GPU, 40 CPU), 40W for the Bluray, RAM, etc. stuff.
Do you want the icecaps to melt?
Wii U enables high power PS4 - average usage for a multi-console household would remain static because Wii U uses less energy.
I've been meaning to ask this for quite some time but do these components really draw that much power?
PS4 will set houses on fire. The carbon footprint would be huge.
systemfehler
Member
Nah they do that anyway. Seriously if Sony/MS want to compete with the PC section at least for some time they need that sort of power either that or pull something completly new out of the box. We talk about 2013/14 - launching in that time "Wii-U" style might work for MS but certainly not Sony.
Probably not - I left that figure as a "buffer" for everything I don't know.
PS1, PS2. They were the weakest of the bunch during their most successful years.
Makes sense, thanks.
jeff_rigby
Banned
It's not going to be Samara but may use the core building blocks that Samara uses. If Jaguar CPUs then it's Kabini or Samara with Samara having the later design using GNB and it looks like a easy to pull out and use building block.
I'm talking the existing building blocks that AMD can use to produce a custom Thebe chipset for PS4.
Roadmap for 20nm from 2011, looks like 20nm volume production would start in the 1st quarter of 2013. http://www.brightsideofnews.com/new...-gate-first-technology-for-20nm.aspx?pageid=1
http://www.dailytech.com/AMD+Concedes+DieShrink+Race+to+Intel+Considers+ARM+Cores/article23931.htm said:
Kaveri and Pennar are rivers in India.http://en.wikipedia.org/wiki/Penner_River said:
Samara is a river in Russia and may indicate a new naming convention for a new series APU due to a different GPU series.
Weren't they more powerful than their sega competitors?
systemfehler
Member
The PS2 was not the most powerfull console but had some areas where it did pretty good. The problem is that after the Wii/Xbox Sony lost their throne - if they want to get it back it is either via being the superior console or some "addon" that can WOW the people so much that they don't care. They compete with MS, Nintendo and the mobile market at the same time - you need some features (like Bluray, ...) to distinguish yourself from the competition. I doubt a PS 3.5 + tablet would in 2014 will be enough to beat all of them.
Hint: A underpowered console with a tablet didn't WOW me...
You know or you guess?
Fair enough. And I don't think it will be a mobile chip either. You're better off going with a last generation cheaper cpu than a mobile apu.
The performance is low, and I mean very low.
Saturn and dreamcast don't count?
I'm just talking about the consoles,he was referring to how PS1 and PS2 were the weakest consoles of their generation even then the difference wasn't that big and most game development was done on them and then ported to others.The competition at least in the PS2 gen launched much later too and both the PS1 and the PS2 were huge upgrades over the then last gen consoles unlike the wiiu.
From what I understand, even a decent gap in theoretical FLOPs would produce minimal differences. The mass market didn't care enough about the differences between the Xbox and ps2 multi-platform games to effect console or software sales. Much like how few people care about the differences we find with PS360 multi-platform games.
Depending on how true early rumors may be, if they are providing more power in the box, it may be to compensate for the larger overhead by the OS.
Another possible reason to provide more power might be to push your platform to be the lead console for development. Whether or not this is the case, I don't see many 3rd party titles exclusively appearing on just one platform next Gen.
I don't understand where you come up with this stuff.
Also that 2010 leak is a pointless marketing concept presentation.
In case anyone was still holding out hope for a Cell+, this was tweeted by a developer:
lol Well that's that.
Who tweeted this?In case anyone was still holding out hope for a Cell+, this was tweeted by a developer: