Predict: The Next Generation Console Tech
- Thread starter Acert93
- Start date
- Status
IllusionistK
Newcomer
28nm AMD GPUs shipping for revenue
I have an open question to ask: Why are we assuming that TDP won't increase exponentially as it did from previous generations. It seems too arbitrary that we are limiting power consumption to ~200w. Why not 300w, 400w, or 500w?
Obviously cost is an issue but my rationale is that perhaps the manufacturers are willing to take a big loss again rather than try to break even as early as possible to better differentiate themselves from the current generation.
A 30x increase in compute != 30x increase in visual fidelity. Crytek made similar statements not too long ago.
------------------------
http://vr-zone.com/articles/amd-28nm-gpus-shipping-for-revenue/14246.html
28/32nm is the bare minimum... I just can't see 40/45nm for GPU/CPU in 2013 when the relatively(to Intel) slow TSMC will have 20nm silicon then.
Perhaps Microsoft will go the big-die(as opposed to small-die 28/32nm) route for 28nm/32nm in hopes of a 2014 20nm die shrink to cut costs.
I have an open question to ask: Why are we assuming that TDP won't increase exponentially as it did from previous generations. It seems too arbitrary that we are limiting power consumption to ~200w. Why not 300w, 400w, or 500w?
Obviously cost is an issue but my rationale is that perhaps the manufacturers are willing to take a big loss again rather than try to break even as early as possible to better differentiate themselves from the current generation.
A 30x increase in compute != 30x increase in visual fidelity. Crytek made similar statements not too long ago.
------------------------
http://vr-zone.com/articles/amd-28nm-gpus-shipping-for-revenue/14246.html
28/32nm is the bare minimum... I just can't see 40/45nm for GPU/CPU in 2013 when the relatively(to Intel) slow TSMC will have 20nm silicon then.
Perhaps Microsoft will go the big-die(as opposed to small-die 28/32nm) route for 28nm/32nm in hopes of a 2014 20nm die shrink to cut costs.
No, I'm simply suggesting the heatsink can easily be bigger in a case the size of the original xbox360 if you put it parallel with the large side of it (instead of putting it on the CPU/GPU and blowing the air out the bottom through a funnel).
Because cost goes up with TDP. Most people believe they won't even launch a loss leader next gen. I'm not quite that cynical myself, but I don't expect anything competing with the very high end of PCs anymore.
And yet IBM confirmed the CPU for the WiiU will be (40/45nm). Then again the Wii-U will be more of a current-gen and not a next-gen console.
Not sure, but the Nanowick tech would be a heat pipe design similar to what the SEGA Dreamcast had but more effective.
I'm more intrested in this new "design boss" improving the gamepad. Cut a deal with SEGA for dpad design (Saturn dpad) or just outright buy their console patents. Then get better haptic feedback and lower power consumption by using Immersions Touchsense rumble technology. Since gamepads are wireless the less power it drains the better and it the feedback is improved it just seems like a win-win.
IllusionistK
Newcomer
I would be content with "Performance"-tier PC hardware for consoles in 2013.
However, I'm still holding out for a customized, bleeding-edge beast. We all want it.
Knowing something as simple as the TDP would be extremely valuable to us.
Indeed.
anexanhume
Veteran
The laptop cost is somewhat immaterial. For instance, what Microsoft did with the Xbox 360 was buy the rights to the Xenos design, ATI/AMD only gets royalties. Thus, microsoft's cost is pure: the fab cost plus a small royalty. They don't need to make a huge markup on the fab cost because they plan to make it up in software, accessories and services.
The difference between a robust solution for cooling and a meager one can be just a few dollars in parts. Those dollars can add up here and there, but if it is necessary, it is necessary.
Bad example IMO. The CPUs tend not be to the most stressed components on these laptops. A more apt comparison is something like the x7200, which uses a 130W desktop processor and is kept cool just fine.
That is that laptop, one with two GPUs and a huge CPU overclock!
And just look how little of it's overall space the M18x has for cooling.
http://www.hardwareheaven.com/reviewimages/alienware-m18x/alienware-m18x_inside.jpg
A console does not need 2 HDDs, a big battery and it can be much thicker than a laptop (M18x is 54mm with the lid closed).
Last edited by a moderator:
I'm agree with many here.
My bet for next gen is something like 1408 SIMD/streamprocessors as a future Radeon HD 7850 (effective performance of a 5870?), As was said by many here, today we have the 6990M (1120 streans/715MHz) under 40nm dissipating 75 watts in relatively small size of notebook (with many things like large battery etc) may be possible place with good cooling system even a gpu 100 watts.
(a 1408/SIMD/HD7850/6950 gpu could reach new highs on a closed box console)
My bet for next gen is something like 1408 SIMD/streamprocessors as a future Radeon HD 7850 (effective performance of a 5870?), As was said by many here, today we have the 6990M (1120 streans/715MHz) under 40nm dissipating 75 watts in relatively small size of notebook (with many things like large battery etc) may be possible place with good cooling system even a gpu 100 watts.
(a 1408/SIMD/HD7850/6950 gpu could reach new highs on a closed box console)
Last edited by a moderator:
$4,224 is quite a lot of money.
I think there are plenty or reasons to keep the TDP down and 200 W and lower is probably where we should have our expectations.
Cooling tech and ACDC-converters also cost money and don´t forget that much of the RROD debacle came from problems with thermal tension.
$4,224 is quite a lot of money.
No one cares. It's not what the discussion was about, but the laptop contains a lot of things no one expects in a console such as a full version of windows, dual discrete graphics parts, a keyboard and 18" LCD panel integrated into a package that you can use on the move as well as a significant profit margin for dell.
A problem that could have easily been alleviated by some engineering and a few dollars of cooling hardware. I really don't know that they'll exceed 200W, but I don't think cooling is really the roadblock in a reasonably sized box especially if they can find a way to ditch optical as its nearly 20% of the boxes (I expect I could use a 360 box and throw in 300watts of PC hardware without too much problem). Cost is a much bigger consideration for the hardware that's going to produce that heat.
Looks noisy. Dunno if the console companies are going to consider the noise profile along with the heat profile, but there's going to be some point where they decide their box is too noisy if they just put lots of small fans in there. PS3's cooling solution looked well engineered and was pretty costly and space consuming. As cooling tech hasn't advanced a massive amount (or any amount beyond fans and heat sinks), perhaps the best cooling we can hope for is something the same as PS3. In which case whatever chips are used need to fit the thermal output of PS3.
Unless PS3's cooling solution wasn't very optimal and better engineering can deal with higher temperatures in the same space.
We could think or imagine some things.
1 - If Manufacturers will follow the same level of development that had been hit with ps3 and x360 comparison with their predecessors, further aggravating the fact that in 2012/2013 will be ps360 with 6 to 8 years as a console solution's supposed "high-end" for the market.
a) if so we may see next gen GPU with powerful and high-end cooling solutions (non-exotic and extremely expensive) to launch-date and in this case Manufacturers will benefit from some re-engineering(die shrink to 20nm,revision PCB) the same level as the ps3 in less than one year as occurred the ps3 60GB to 40GB (my ps3 have date fab october/2007) without any ps2 hardware,SD cards slots,only 2 USB,Without Super Companion Chip etc;
b) if MS and Sony come with hardware at this level (10 * or more than ps360) and extremely unwilling to modify their internal systems to maximize production scale and perhaps longer come with closed box in sizes large than seen previously.
2 - If coming with "medium range" or even "low range" betting on hardware cycles (from 5/6 years to 4 years) under its "durability on the market" to face the universe tablet / cell phone.
a) In this case perhaps Manufacturers like Sony and MS think of a solution not so far for cooling and too close in performance (and size) to what we are waiting (speculating gpu less than HD 4870 and 3 to 4 core cpu powerpc A2/PPU like) for the wii U;
b) Even in the event of hardware medium range or low range with SoC like maybe they might not need more powerful cooling systems we have seen in ps360 current.
There are many "if", but perhaps have pronounced longevity of the current generation face with the latest developments of high-end PC hardware coupled with the growing market tablet / cellphone games ,makes Manufacturers think and rethink each step to follow surprising us favorably...(sorry all ...i'm praying for that! Cause will be a nightmare games predominantly casual like).
(imagine/dreaming mode off)
1 - If Manufacturers will follow the same level of development that had been hit with ps3 and x360 comparison with their predecessors, further aggravating the fact that in 2012/2013 will be ps360 with 6 to 8 years as a console solution's supposed "high-end" for the market.
a) if so we may see next gen GPU with powerful and high-end cooling solutions (non-exotic and extremely expensive) to launch-date and in this case Manufacturers will benefit from some re-engineering(die shrink to 20nm,revision PCB) the same level as the ps3 in less than one year as occurred the ps3 60GB to 40GB (my ps3 have date fab october/2007) without any ps2 hardware,SD cards slots,only 2 USB,Without Super Companion Chip etc;
b) if MS and Sony come with hardware at this level (10 * or more than ps360) and extremely unwilling to modify their internal systems to maximize production scale and perhaps longer come with closed box in sizes large than seen previously.
2 - If coming with "medium range" or even "low range" betting on hardware cycles (from 5/6 years to 4 years) under its "durability on the market" to face the universe tablet / cell phone.
a) In this case perhaps Manufacturers like Sony and MS think of a solution not so far for cooling and too close in performance (and size) to what we are waiting (speculating gpu less than HD 4870 and 3 to 4 core cpu powerpc A2/PPU like) for the wii U;
b) Even in the event of hardware medium range or low range with SoC like maybe they might not need more powerful cooling systems we have seen in ps360 current.
There are many "if", but perhaps have pronounced longevity of the current generation face with the latest developments of high-end PC hardware coupled with the growing market tablet / cellphone games ,makes Manufacturers think and rethink each step to follow surprising us favorably...(sorry all ...i'm praying for that! Cause will be a nightmare games predominantly casual like).
(imagine/dreaming mode off)
Last edited by a moderator:
You could fill a console up with good old 5000 rpm delta fans and it would only match the noise of the original xbox360 DVD drive ...
Given the experience of the current generation of consoles and the way the world is changing I think we ought to shape our projections around these three obvious developments.
1. The console will have low standby power consumption. This is likely to be mandated by governments around the world regardless.
2. The console will not be as noisy as the original 360s.
3. Improvements to user experience will take precedence over increase performance.
Improvements to user experiences cannot be wasted until it hits the point of severe diminishing returns whereas improvements to performance can be wasted quite simply with additional graphics effects slapped on with little to no noticeable difference from the perspective of the end user. What incentive is there for a console manufacturer to have the most bleeding edge hardware anymore? They don't make profit off the tears of developers!
I say the majority of the important technology will be directed towards the interface, user experience and the software that drives it. The Xbox 360 currently, the Wii and the Wii U all have a significant proportion of their BOM directed towards the user interface, a much higher proportion than was traditional in the past. Regardless of the price this mere fact means that the hardware inside the console itself will not be as powerful or expensive relative to whatever final price is set.
The thing which really complicates any projection is the software/hardware mix. Whilst for instance having a flash based cache or keeping the memory active to quickly restart a game from standby might divert money away from other areas, the same could also be achieved with pre-emptive loading using software facial recognition and probability algorithms which could cause the console to preemptively load the top 3 most likely games which causes the game to start nearly instantly after you've gotten around to putting your coffee on the table. No hardware is faster than anticipation.
If we take a $400 console and if we expect that ~60% of the cost is going to be non performance orientated bits n pieces such as interfaces, margins, boxes, shipping, royalties and drives etc then it is probably best that we take something pretty similar and high margin as a basis for what performance we'll get. The best proxy we have IMO is a video card of roughly the same value. We can divide the silicon budget over various components and come up with a rough transistor count and costing. If a $400 video card in 2013 has say 4GB RAM and 413mm^2 GPU then a console built on the same process likely has around the same quantity of silicon and memory as they are both similarly embedded platforms.
1. The console will have low standby power consumption. This is likely to be mandated by governments around the world regardless.
2. The console will not be as noisy as the original 360s.
3. Improvements to user experience will take precedence over increase performance.
Improvements to user experiences cannot be wasted until it hits the point of severe diminishing returns whereas improvements to performance can be wasted quite simply with additional graphics effects slapped on with little to no noticeable difference from the perspective of the end user. What incentive is there for a console manufacturer to have the most bleeding edge hardware anymore? They don't make profit off the tears of developers!
I say the majority of the important technology will be directed towards the interface, user experience and the software that drives it. The Xbox 360 currently, the Wii and the Wii U all have a significant proportion of their BOM directed towards the user interface, a much higher proportion than was traditional in the past. Regardless of the price this mere fact means that the hardware inside the console itself will not be as powerful or expensive relative to whatever final price is set.
The thing which really complicates any projection is the software/hardware mix. Whilst for instance having a flash based cache or keeping the memory active to quickly restart a game from standby might divert money away from other areas, the same could also be achieved with pre-emptive loading using software facial recognition and probability algorithms which could cause the console to preemptively load the top 3 most likely games which causes the game to start nearly instantly after you've gotten around to putting your coffee on the table. No hardware is faster than anticipation.
If we take a $400 console and if we expect that ~60% of the cost is going to be non performance orientated bits n pieces such as interfaces, margins, boxes, shipping, royalties and drives etc then it is probably best that we take something pretty similar and high margin as a basis for what performance we'll get. The best proxy we have IMO is a video card of roughly the same value. We can divide the silicon budget over various components and come up with a rough transistor count and costing. If a $400 video card in 2013 has say 4GB RAM and 413mm^2 GPU then a console built on the same process likely has around the same quantity of silicon and memory as they are both similarly embedded platforms.
And right now, I would say MS is winning that battle. They have since the system came out. The 360 is just so easy to use, there is so much available on Live even outside of gaming. The addition of Kinect is obviously working quite well with consumers too, so I think Sony pretty much is on the tail end, countering whatever MS is doing.
If the focus is this kind of route, where system performance is going to take the back seat, then I think the 100W power consumption goal is reasonable for giving devs much more powerful hardware without creating a toaster oven prone to breakage. Then that would be a what, 125-150 TDP worse case?
Last edited by a moderator:
There is couple of big differences between laptop cooling and console cooling that haven't been discussed yet.
1) Laptop can downclock if it becomes too hot, console always need to be able to work at peak performance
2) Combine 1 and worst conditions and Do Not Fail(hot summer day, high ambient temperatures)
I wouldn't expect what has been stuffed inside laptop and what seems to work there to function on a console reliably when you combine 1&2 from above. Especially not so when you run the console at peak performance for 6h+ gaming session...
1) Laptop can downclock if it becomes too hot, console always need to be able to work at peak performance
2) Combine 1 and worst conditions and Do Not Fail(hot summer day, high ambient temperatures)
I wouldn't expect what has been stuffed inside laptop and what seems to work there to function on a console reliably when you combine 1&2 from above. Especially not so when you run the console at peak performance for 6h+ gaming session...
Last edited by a moderator:
Microsoft set the current precedent with their $100 Kinect. IIRC the Move is probably the next most expensive controller and I expect the Wii U to set the future precedent for how expensive/complex a controller can be. The success of Live/Wii/Kinect and their profitability suggest that the major focus is going to be software and interface and the hardware had better not get in the way.
I think of the Xbox 360 S and the PS3 Slim to be the true predecessors to the next generation consoles from Microsoft and Sony. Whatever they come out with will be a replacement of the 360 S with Kinect and the PS3 Slim with Move and not the over-costed performance behemoths of the original designs. A good estimate of the Xbox 3 is what you'd consider a clean sheet equivalent design of the Xbox 2 Slim 250GB with Kinect for $399 in 2013 on a mature 28nm process. They ought to be able to slap in 3-4B transistors at a faster clock speed with 4GB of RAM for that kind of money and still make a small margin on it, even with Kinect.
Edit: They're probably looking REALLY hard at the justification for that HDD for what it's worth. I'm sure they have data which shows what % of the HDD space is actually used and are weighting it against using flash as a substitute. If games are 15+ GB and therefore impractical to download then they may consider it unworthwhile to pursue and if internet speeds improve to the point where that 15GB game can be downloaded then it is quite likely it can also be streamed.
Last edited by a moderator:
- Status
Similar threads
- Locked
