Predict: The Next Generation Console Tech

[makattack]

Changing course slightly, I think this is the future: standard gaming platforms vs. proprietary console platforms. Perhaps it's all server (cloud) based, or a common client platform -- but there are indications this could well be the direction we're headed.

http://www.gamasutra.com/view/news/...l_Not_Be_Tied_To_Platforms_In_Near_Future.php

 

[aaronspink]

Changing course slightly, I think this is the future: standard gaming platforms vs. proprietary console platforms. Perhaps it's all server (cloud) based, or a common client platform -- but there are indications this could well be the direction we're headed.

http://www.gamasutra.com/view/news/...l_Not_Be_Tied_To_Platforms_In_Near_Future.php

So I have this great piece of real estate, lots of traffic, waterfront views, its a steal. So make me an offer and I've sure we can get this all settled. We can even help arrange any bridge loans that are required.

 

[Duck Dodgers]

Changing course slightly, I think this is the future: standard gaming platforms vs. proprietary console platforms. Perhaps it's all server (cloud) based, or a common client platform -- but there are indications this could well be the direction we're headed.

http://www.gamasutra.com/view/news/...l_Not_Be_Tied_To_Platforms_In_Near_Future.php

Heh I said this as much nearly two years ago, albeit in a trollish fashion. Apologies.

 

[MfA]

The SCC from Intel is interesting ... a nice quote from one of the presentation :

We believe software managed coherency on non-coherent many-core is the future trend

A row brewing at Intel? This is clearly a different paradigm than the Larrabee team, which was already working on complex scheme to scale snooping for future Larrabees ... the SCC team is course clearly right, a mesh is inherently better scaling than the hierarchical ring structures with hidden directories in between.

 

[rpg.314]

The SCC from Intel is interesting ... a nice quote from one of the presentation :

Linky?

A row brewing at Intel? This is clearly a different paradigm than the Larrabee team, which was already working on complex scheme to scale snooping for future Larrabees ... the SCC team is course clearly right, a mesh is inherently better scaling than the hierarchical ring structures with hidden directories in between.

Good riddance, I'd say. As long as the message passing is swept under a carpet prettier than MPI, I don't care.

 

[nAo]

Linky?

http://techresearch.intel.com/UserFiles/en-us/File/terascale/SCC_Sympossium_Feb212010_FINAL-A.pdf

 

[rpg.314]

http://techresearch.intel.com/UserFiles/en-us/File/terascale/SCC_Sympossium_Feb212010_FINAL-A.pdf

THNX.

 

[gongo]

Will Sandy Bridge/Bulldozer make a good console CPU?

In my research for a new PC, i saw things that made me questioned...the need for special custom silicon in next gen consoles' CPU...im a lover of CELL and would like to see its return to PS4...but what if next gen just use a "normal"...x86 CPU like a Sandy Bridge...22nm 3.4ghz, 6 cores..not exactly hot by 2012 3Q.

Sandy Bridge high end parts to replace Bloomfield comes next year 2Q...32nm, 6-8 cores, integrated PCI-e. 2012 is enough to see a tick-tock refresh for consoles to use the laptop bins. Sandy Bridge promises twice the perf of Nahelem FPU/Multimedia IIRC...sounds good? Bulldozer also sounds good...very scalable...but i heard AMD is choosing to strengthen the Integer side...wonder how it will affect games....but MS is more likely to choose a complete AMD solution?

I looked at all the latest games benched...and they becomes GPU limited at 1080p with 2XAA..most quad cores arent even half utilised......i dont expect nextgen consoles to compete with high end GPU in bandwidth/ram/fillrate? There should be a lot of processing muscle in today's i7 cpus....to tap it...

I heard stories about PS3/360 "custom" CPU...and they are not good...PS3 at least is able to calculate certain specialized tasks very fast...both CPUs were still very slow at what you expected from them...like Pentium4 dual cores slow in 2005/6...

Other thoughts like, a console has to meet requirements of a small enclosure, quiet, low heat....making smaller processors is getting harder...Intel is ahead of the game here..all parts must come together...something like a HTPC/Laptop build works quite well.

 

[TheWretched]

Well, you are right in that TODAYS games might be GPU limited at those resolutions... but we are talking game development that starts today, with the expectation of 6 cores in 2 years. And a console should last 5 years or more, so there should be some headroom.

But I am impressed at how much faster the Core i3 already is, compared to my X2 (330m at 2.13Ghz vs. 2.2Ghz) in stuff like x264 or even PCSX2 (which runs nearly full speed on my laptop now )

Thing is with consoles compared to general purpose PCs is, that the use case is very limited within games. Seeing that PS3 already could run Linux acceptable (mostly limited by RAM constrains and no GPU acceleration) and even a webbrowser, there's no need for more power there, since the general steps up in CPU design will alleviate the contrains there, especially more RAM.

CELL in PS3 is, imho, a good choice, as can be seen with many first party titles. I just finished God of War 3, and that game is just on another level compared to many other current multiplatform titles. And this is mostly due to CELL, since RSX is quite limited compared to XeGPU.

Going forward, I doubt Sony wants to drop BC again, like they did with PS3... So I guess they either add something like Spurs as a DSP (the PS1 chip was once used as IO processor in PS2, until it was emulated in software and designed away) and for BC plus a more general purpose CPU compared to CELL itself. That way, Sony could probably put the die into one package with the CPU at a later stage to save costs, too. And by 2012 or whenever a new Playstation hits, CELL will be quite small too.

This way, Nvidia "must" stay too, for BC reasons. I have no idea how feasible it is to emulate something like RSX on another GPU, but given that Sony wasn't able to emulate the GS yet, I don't see them doing it with RSX either, unless they use another Nvidia chip.

As it stands now, Nvidia has the faster Dx11 chip too, but at the price of high temperatures and power usage. But there's still a long way to go until we see new consoles from either company.

 

[corduroygt]

It would be a bad cpu because it'd be too expensive. Intel has the highest price/mm2 ratio chips out there. Also, having designs tied to advanced manufacturing processes that only intel can deliver is not very good, since you'd be at their mercy.

 

[Grall]

Tbh, Intel's pure CMOS performance/mm2 is also highest in the business, and they're typically a full node ahead of everyone else too. TSMC's 40nm process still isn't up to snuff compared to Intel's, and intel is already working towards migrating away from 32nm...

The reason they're also the most expensive is because they're market leaders, and can charge whatever they like pretty much. There's nothing inherently uber expensive about their product - it's just priced that way for business reasons. If Intel wanted to get into console space again they'd offer competitive prices, and whomever they team up with would automatically get free world-class, bleeding edge process shrinks on a timely and reliable basis.

You could be doing much MUCH worse than that...!

 

[V3]

Better spent the budget on GPU instead of CPU. Sandy Bridge and Bulldozer based CPU will be overkilled for console. Cell at 4+GHz with all the SPUs will be good enough for PS4. What it needs is a good GPU or two. Next gen will be either displacement mapping or voxels. Better pick GPU that's good at them rather than worrying about CPU architecture.

 

[function]

Bobcat or IBM again is my guess.

Can't see console makers wanting something with an integrated memory controller, or something where they can't hunt around for the cheapest fab as the console grows older.

 

[corduroygt]

If Intel wanted to get into console space again they'd offer competitive prices, and whomever they team up with would automatically get free world-class, bleeding edge process shrinks on a timely and reliable basis.

You could be doing much MUCH worse than that...!

Why would Intel want to make low-margin console cpu's in their bleeding edge fabs (currently they're 32nm), when they can sell every chip they make with that process with fat profit margins? It makes no business sense for Intel. If anything, they'd be willing to use older fabs (65nm or larger) that they currently use for chipsets, etc. since they're more likely to have excess capacity there.

Oh and this thread should be merged with the next gen console tech thread.

 

[Shifty Geezer]

I concur, and have merged.

 

[darkblu]

In my research for a new PC, i saw things that made me questioned...the need for special custom silicon in next gen consoles' CPU...im a lover of CELL and would like to see its return to PS4...but what if next gen just use a "normal"...x86 CPU like a Sandy Bridge...22nm 3.4ghz, 6 cores..not exactly hot by 2012 3Q.

I would not worry that much that PS4 would not use CELL - sony made too big of an investment with the CELL to just discard it freely. It would make much more business sense for them to actually provide the software tools they did not have/were late with on the PS3 from the get go, and bump up the quantitative characteristics of the CELL in the PS4. Double the SPE's, offer middleware similar to DICE's frostbyte engine so developers who don't feel like designing a packet-based pipeline wouldn't have to rediscover the wheel, let alone feel intimidated by the platform - voila, there's your PS4 core architecture, the done-right successor to the, well, revolutionary PS3.

 

[PalmiNio]

Why would Intel want to make low-margin console cpu's in their bleeding edge fabs (currently they're 32nm), when they can sell every chip they make with that process with fat profit margins? It makes no business sense for Intel. If anything, they'd be willing to use older fabs (65nm or larger) that they currently use for chipsets, etc. since they're more likely to have excess capacity there.

Oh and this thread should be merged with the next gen console tech thread.

Why?

1. If you do not do it, you competitior will. Intel do not need it but why invite the competition to stick around in the game?

2. Long term investment. Console sales grow over a long period of time and you have no need to make the chip faster over time, just cheaper.

Sure Intel does not need to do it, but being busted by IBM on contracts to all three console producers must sting. In the long term they can not just stock with CPU for home computer, specielly not when competition grows with boom of mobile devices.

Hard thing is to provide som sort of BC to any of the consoles.

 

[Shifty Geezer]

In the long term they can not just stock with CPU for home computer, specielly not when competition grows with boom of mobile devices.

I think the markets for PCs, servers and netbook+laptops are far larger and more lucrative than the cut-throat pricing of the consoles.

 

[MfA]

Intel will play ball if Bobcat performs well enough to be an option for the next generation, even if the profit margins aren't up to their usual standards.

 

[corduroygt]

Shifty nailed it, millions of computers sold every year (desktops, notebooks, netbooks, server) is a much bigger AND higher profit margin than consoles. Especially when their best process node is capacity constrained, hence the 45 nm IGP paired with the 32nm Arrandale. If Intel goes in the console business, they won't be using their best process nodes, which sort of defeats the purpose of choosing Intel in the first place.

The biggest semiconductor markets are computers, servers, then I'd guess mobile devices and GPU's, not consoles. Intel would try to compete in those markets first, especially the GPU market since GPGPU's will be threatening to take over CPU's in the next 10 years.