Predict: The Next Generation Console Tech

Status
Not open for further replies.
And taking that in to account I still think a high spec PC from 2008 will be in the ball park of a 2011 console. I know we would all like the consoles to BATTER PC offerings, but its not likely to happen.

If a high end PC at the end of 2008 with SLI'd top end cards can run Crysis in 1080p or there abouts in high detail with 2xAA and 16x AF then that for me is about the right target performance for a next gen console, give or take.

Lets do the numbers. In those three years, we will hopefully have progressed one lithographic process node. Two SLI´d top end cards could concievably be condensed to one top end card by the time the console would ship. (Note that consoles don't launch at the bleeding edge of lithographic technology, they need to be able to hit their launch dates, and ship in volume.)

So - do you think that the console makers will be prepared to launch with the equivalent of a top end PC graphics card? Would that fit within their cost and power draw constraints?

It is within the realm of possibility, but is it likely? I'd say that it is an outer bound. If, having learned from their woes this generation, Sony and Microsoft decide to fit within a tighter power (and cost) envelope, expectations need to be adjusted down from there.

Personally, I'm more curious about what will be possible within a 10-20W envelope, where the Wii launched this generation.
 
If a high end PC at the end of 2008 with SLI'd top end cards can run Crysis in 1080p or there abouts in high detail with 2xAA and 16x AF then that for me is about the right target performance for a next gen console, give or take.
Considering that both consoles will go for a 5 years life-cycle minimun , this will give us a xbox3 at Q4-2010 and a PS4 at Q4-2011.

For next xbox:
To say that a G200 from Q4-2008 will have the same graphical capabilites with xbox3 at Q4-2010, is like saying that a FX 5950 is more or less equal to Xenos :oops:
SLI or not, they will not be in the same league*.

For PS4:
Q4-2008 G200
Q4-2009 G300
Q4-2010 G400
Q4-2011 G500 (A derivative of G500 as PS4 GPU*)

( *Without even mentioning the high possibilities for dual GPUs :p )
 
Considering that both consoles will go for a 5 years life-cycle minimun , this will give us a xbox3 at Q4-2010 and a PS4 at Q4-2011.

For next xbox:
To say that a G200 from Q4-2008 will have the same graphical capabilites with xbox3 at Q4-2010, is like saying that a FX 5950 is more or less equal to Xenos :oops:
SLI or not, they will not be in the same league*.)

Not really, multi-GPU was not something available back then, it came with the Geforce6 series (and it was in "child state"). And not to forget the 5xxx series from Nvidia was a mess due to the architecture.
Nowdays you have SLI and trippel SLI available, making huge perfomance difference. Then add that a videocard has 2 GPU's and a trippel SLI system would most likely be there.
Price here is not really important right now 'since you pay to play the future'! :p

For PS4:
Q4-2008 G200
Q4-2009 G300
Q4-2010 G400
Q4-2011 G500 (A derivative of G500 as PS4 GPU*)

( *Without even mentioning the high possibilities for dual GPUs :p )

That would most likely make the price of the PS4 sky-high, something the consumers wont like. People want cheap consoles. And about 2 GPU's I would say 2 non-top of the line GPU's for cost, yield and perhaps power draw as best. 2 top of the line GPU's is out of the question cost and heat wise. :smile:
 
Nowdays you have SLI and trippel SLI available, making huge perfomance difference. Then add that a videocard has 2 GPU's and a trippel SLI system would most likely be there.

The dual GPU cards only feature enough connections for one other dual-GPU card. ATM there's no such thing as 6-way SLI. Same goes for Crossfire. 4-way is as high as it gets right now.
 
( *Without even mentioning the high possibilities for dual GPUs :p )

There's zero possibility of dual GPU's in any next gen console. Absolute zero.

I agree though that following historical precident, the GPU in in XB4 in Q4 2010 should be significantly more powerful than GT200.

What someare suggesting though is that the next gen consoles won't go as technologically far as they have done in the past. Rather than going for top end PC like performance they may only aim at mid range PC like performance for heat/power/cost reasons.

2x GT200 in SLI might match mid range performance of Q4 2010.
 
( *Without even mentioning the high possibilities for dual GPUs :p )
You want to bet on that? :)
However, you're almost certainly right that the GPU won't use 2008 level technology - It will either be an extension of the existing architecture (for compatibility reasons) or, more likely, use whatever goes for current around launch.
 
Nebula said:
Not really, multi-GPU was not something available back then, it came with the Geforce6 series (and it was in "child state"). And not to forget the 5xxx series from Nvidia was a mess due to the architecture.
Yet it doesn't change anything in my example. If you want you can compare 6800 in sli vs R600-G80: still they are not in the same league.

Entropy said:
You want to bet on that?
However, you're almost certainly right that the GPU won't use 2008 level technology - It will either be an extension of the existing architecture (for compatibility reasons) or, more likely, use whatever goes for current around launch.
If you think about it, its more likely that both consoles will skip the DX10 league of GPUs (Q4 2006 - Q4 2010: it makes sense).
So,it would be interesting to predict the DX11 technical specs their next GPUs will have to support. ( Maybe multicore GPU support? :LOL:)
 
You want to bet on that? :)
However, you're almost certainly right that the GPU won't use 2008 level technology - It will either be an extension of the existing architecture (for compatibility reasons) or, more likely, use whatever goes for current around launch.

I wasnt talking about architecture per se I was talking about performance metrics. I think it will be some sort of DX11 style architecture.

WRT performance a top sli config late 2008 maybe under spec, but with the route past 22nm / 2012 being less certain than before. They may not be able to afford to push the envelope as hard as they did with PS2/PS3 and to a certain extent xbox360.

WII / PS3 have demonstrated the importance of a lower price point. I dont think they will relsease such expensive consoles again. We are back to the £299 launch prices of yester year. Factoring inflation thats cheap for 2012. How do you suggest they will achieve that if they put some souped up top end card from 2010 it just wont scale down in time to fit the price bracket.

Unfortunately I dont think the jumpo will be as high as people hope. But it will still be enough for excellent performance at 1080p. Im looking forward to it.
 
Yet it doesn't change anything in my example. If you want you can compare 6800 in sli vs R600-G80: still they are not in the same league.

Not in the same league true, but then G80/R600 is PC tech thats ahead of the rest of GPU's available (except the new ones if they are out now).RSX/Xenos cant hold it against them.

The dual GPU cards only feature enough connections for one other dual-GPU card. ATM there's no such thing as 6-way SLI. Same goes for Crossfire. 4-way is as high as it gets right now.

I see, well perhaps it will come out by then or '09. Neverthless 4-way SLI should do it!
 
If you think about it, its more likely that both consoles will skip the DX10 league of GPUs (Q4 2006 - Q4 2010: it makes sense).
So,it would be interesting to predict the DX11 technical specs their next GPUs will have to support. ( Maybe multicore GPU support? :LOL:)
Prior DX versions feature-sets have been limited by the hardware possibilities of the time. There was no point in 65536 instruction pixel shader for DX7 when the hardware couldn't be designed to run them effectively. As hardware improves, software limits end up being a constriction on GPU functionality. We see hacks to fit problems onto GPUs, in the form of GPGPU. DX11 should be basically free coding. There'll be no real need for software restrictions. The hardware would be a fully programmable as a vector float processing array, used in effect as a software renderer, turning its hand to whatever techniques the developers decide to use. Software API's will offer access through existing common interfaces, or you can throw them to the wind and write everything low-level.
 
I see, well perhaps it will come out by then or '09. Neverthless 4-way SLI should do it!

Oh god, 6 way SLI? Please nooooo! ;)

Seriously though, I don;t like SLI, even two way. I would much prefer a move back to single GPU's at the high end only. For one its cheaper for those who want the best and two, it reduces the gap between the low and high end that devs keep complaining about.

It also reduces the percieved price of a "top end" gaming rig for all those that claim PC gaming is too expensive.
 
Can't say I so much mind SLI - It's just SLI in the gaming market that annoys the piss out of me. It doesn't belong there - It belongs in the arena of large-scale, high-complexity simulations, but current SLI systems aren't really geared for that, either (Too limmited).

IMO, it promotes a bad development practice in the gaming space (games that at max settings, run at 20 FPS on single GPU?!? WTH!)
 
Interface over Graphics?

I expect the next iteration won't until 2009. But based on the huge success of the Wii, perhaps the adances will be in interface rather than graphics.

I talked about this very subject on my show this week. Listen here.

Growing Up Geek
Geek Entertainment and Nostalgia
 
My predictions:

The next generation won't be as expensive as the last one, due to the success of the low priced Wii and lower shrink potential beyond 22 nm. So i expect the chips to be only half as "big" as the last ones (in mm^2).

PS4:

Release in H2/2012, max. 400$, 32 nm (while Intel is already at 22 nm :rolleyes:).

Cell 2 with 2 PPEs and 32 SPEs (but only ~30 used), 4-5 GHz.
It should have around 1 billion transistors and half the area (in 32 nm) of the 90 nm Cell 1. Early IBM roadmaps show the Cell 2 for 2010, but i guess the chip will be a bit late like his predecessors so developers should be able to get their hands on Cell2 PS4 Dev Kits in H2/2011.
GPU could be anything (~ 1 billion transistors). No idea, even a second Cell2 could be possible :)

Xbox 720:

Release in Q4/2011, max. 400$, 45 nm (maybe, maybe 32 nm for the CPU).

8-10 core Xenon2 CPU, 4-5 GHz.
Xenos2 GPU with 32 MB eDRAM (for 720p/4xAA without tiling).

PS4 & Xbox 720:

2-3 GB unified RAM (GDDR5 or XDR3 RAM)
8x Blu-ray (50-100 GB)
250-320 GB 2,5" HDD for the "normal" SKU (max. 2,5" HDD available in 2012 will be 1.5 - 2 GB).
Maybe some sort of High-speed SSD "Cache" (i.e. ONFI3, ~32 GB, >400 MB/s read), in that case the HDD could be optional or only in the premium SKU.

I know, not very optimistic. But i don't think Sony and Microsoft will invest this much for the next generation and they won't go uber-high-end and high-priced. I think they will go the save and secure way (evolution instead of revolution).
 
My predictions:

The next generation won't be as expensive as the last one, due to the success of the low priced Wii and lower shrink potential beyond 22 nm. So i expect the chips to be only half as "big" as the last ones (in mm^2).

I know, not very optimistic. But i don't think Sony and Microsoft will invest this much for the next generation and they won't go uber-high-end and high-priced. I think they will go the save and secure way (evolution instead of revolution).

I disagree, I think the next gen BR drive will be much cheaper, therefore the extra cost BR added to the PS3 this go around can be absorbed into making sure the CPU, GPU and RAM are up to snuff in PS4. I also don't think they will push more then 250GB for storage on a HDD; call me crazy but I think even MS knows that the bandwidth required for downloading HD content in any vast amounts is going to be extremely limited even by 2011 (3 years away). MS should have learned from Sony and should incorporate a standard HDD format that is easily upgradeable. I also think both companies should release the first Gen PS4/720's with standard 3.5HDD and wait until the console is small enough to incorporate the smaller drives, possibly make them hot swappable with some form of Fair Use DRM that allows them to just plug it in like an old PS1 memory card.

I think the silicon budget will be just as high next go around as it was this generation. Most of the extra tech inside these machines (Wifi, HDD, Optical Drives) are getting cheaper and cheaper with no real need to upgrade them any time soon. Adding 802.11n wifi a larger hard drive and possibly a faster BR drive should not cost any more in 3 years as what the current tech inside the PS3 is today. (BOM is less today then it was at launch) Since no new technology (as in what DVD was for Xbox/PS2 and what BR was for PS3, built in wifi, built in HDD-PS3, wireless controllers, bluetooth etc etc) will be integrated into these machines it would reason that the extra R&D and BOM could be put forth for greater silicon budgets.

With that said I think the biggest potential problem for the next generation is the continued lack of sufficient ram. I don't want an excessive amount like what many PC games require but 512 this gen no matter how its divided is insufficient for these processors and GPU's. I know it was a significant jump from last gen but in my opinion graphic memory does not scale the same as system memory and that should be accounted for.

Can we sacrifice speed for quantity? I guess thats a question I wont even bother trying to give my opinion on.
 
I disagree, I think the next gen BR drive will be much cheaper, therefore the extra cost BR added to the PS3 this go around can be absorbed into making sure the CPU, GPU and RAM are up to snuff in PS4.
Blu-ray will be pretty much standard by than and much cheaper, but Sony can't launch the next generation with a 600$ price tag. It's clear that the high price tag killed the sales in the first 1,5 years. So yes, they are saving money but they need these savings for an acceptable launch price tag (400$).

I also don't think they will push more then 250GB for storage on a HDD; call me crazy but I think even MS knows that the bandwidth required for downloading HD content in any vast amounts is going to be extremely limited even by 2011 (3 years away). MS should have learned from Sony and should incorporate a standard HDD format that is easily upgradeable. I also think both companies should release the first Gen PS4/720's with standard 3.5HDD and wait until the console is small enough to incorporate the smaller drives, possibly make them hot swappable with some form of Fair Use DRM that allows them to just plug it in like an old PS1 memory card.
I think 250 - 320 GB, depending on the time frame (2011/2012)

I think the silicon budget will be just as high next go around as it was this generation. Most of the extra tech inside these machines (Wifi, HDD, Optical Drives) are getting cheaper and cheaper with no real need to upgrade them any time soon. Adding 802.11n wifi a larger hard drive and possibly a faster BR drive should not cost any more in 3 years as what the current tech inside the PS3 is today. (BOM is less today then it was at launch) Since no new technology (as in what DVD was for Xbox/PS2 and what BR was for PS3, built in wifi, built in HDD-PS3, wireless controllers, bluetooth etc etc) will be integrated into these machines it would reason that the extra R&D and BOM could be put forth for greater silicon budgets.
True, but they need these savings for the 400$ price tag and so that the losses in the first years aren't that big. So the question is if they will spend as much money on the CPU and GPU as for the current generation, or go the Wii way, at least half-way. I think the success of the Wii (with nearly instant profit for nintendo) and the declining High-End PC games market shares will push them to the "cheaper" path. The second big problem is scalability beyond 22 nm. From today's view it is very uncertain that they can scale down their chips in similar periods of time as they have done in the past. If the start in 2010 or early 2011 with 45 nm than that may be a smaller problem, but i strongly believe we won't see the next generation before Q4/2011, more likely in 2012, so 32 nm would be the given starting node. But of course, only time will tell.

With that said I think the biggest potential problem for the next generation is the continued lack of sufficient ram. I don't want an excessive amount like what many PC games require but 512 this gen no matter how its divided is insufficient for these processors and GPU's. I know it was a significant jump from last gen but in my opinion graphic memory does not scale the same as system memory and that should be accounted for.
Sadly, consoles always disappoint in that area. No one wants to believe that they do it again, but in my opinion it will happen. I don't think they will have more than 2 GB unified RAM, if we are very very lucky we get 3 GB (but only if they launch in 2012, no chance in 2010).

Personally, i would like to see "super high-end" next generation consoles (it makes me sad that the low-end Wii is the dominant console this generation). But i just don't think it's realistic.
 
@Mike, a 64/whatever ps2 had to 512mb jump was quite substantial. Theres no reason not to go for minimum 2gb, but possibly 4-8gb. Game designers will always find a way to use ram! They want bigger worlds (Not such a tunnel effect) They are going to be targeting 1080p at a higher detail level than we're used to. Ram will be cheap. Read speeds from media will become a bottleneck, so reading compressed textures into memory would be excellent. 7200rpm on a high density HDD is SLOW, everyone knows that the HDD is one of the major bottlenecks in a P.C. Ram is a way to compensate for it. I bet if you asked a game designer what he'd do with 8gb on a closed system, he'd smile! Not give you a "WTF!? I don't want that!" look.

Actually, I came here to query... What would an all Intel console look like? Or maybe intel+ATI GPU. Would it have a Xeon? Nehalem? Larrabee? Or for AMD's side, you have Fusion and whatever r800/900/infinity. A strictly P/C based console! :)
 
on Intel side I can see a Larrabee-based system, no special need for additional CPU (or GPU), it should do the work of both well enough.
 
@Mike, a 64/whatever ps2 had to 512mb jump was quite substantial. Theres no reason not to go for minimum 2gb, but possibly 4-8gb. Game designers will always find a way to use ram! They want bigger worlds (Not such a tunnel effect) They are going to be targeting 1080p at a higher detail level than we're used to. Ram will be cheap. Read speeds from media will become a bottleneck, so reading compressed textures into memory would be excellent. 7200rpm on a high density HDD is SLOW, everyone knows that the HDD is one of the major bottlenecks in a P.C. Ram is a way to compensate for it. I bet if you asked a game designer what he'd do with 8gb on a closed system, he'd smile! Not give you a "WTF!? I don't want that!" look.
All true, BUT the game designers didn't want 512 MB RAM (for CPU+GPU) in 2006 either, and look what they got. :rolleyes: Also the RAM wouldn't be off the shelves DDR2 RAM, but GDDR6 or something like that.

8 GB would be a dream, but by 2011/2012 that won't even be "normal" for PCs.

Actually, I came here to query... What would an all Intel console look like? Or maybe intel+ATI GPU. Would it have a Xeon? Nehalem? Larrabee? Or for AMD's side, you have Fusion and whatever r800/900/infinity. A strictly P/C based console! :)
2012 we will have the 22 nm Sandy Bridge derivate and third generation Larrabee. Quite possible that we will see some integrated IA++ Larrabee-like units with stacked memory, so a SoC with decent graphics might be possible. Although a separate CPU and GPU will still be much faster. But Intel chips are simply too expensive for consoles.
 
Status
Not open for further replies.
Back
Top