Predict: The Next Generation Console Tech
- Thread starter Acert93
- Start date
- Status
Image quality >>>> Pixel quality
I'm not too say the idea is mad, as far as mad idea are concerned I posted something yesterday that included only 1GB of ram
If memory serves right I believe that you're not the only one who played with the idea, I believe it's been considered when the IBM release its statement about the WiiU.
For the bandwidth requirement I don't know how much is needed. May be somebody could give us information, It would makes sense to me that it scales with the number of render targets, their resolutions, theirs precision, etc. I don't know either if the data flow is constant or a bit burst ( which force you to over over size your bus).
The 360 had 32GB/s and basically I think nobody asked if it set any kind of limitation to the system. Still the link to the shader cores is faster than the link to the main RAM. It's the highest bandwidth connection between 2 piece of silicon we found on any nowadays system. If one is not conservative and consider MS to aim pretty high so to 1080p and fp16 precision I would believe that the requirement for the aforementioned link would be multiplied by 3 (twice as much pixels half of those data would be twice as heavy (the colors)). So if my reasoning is correct we're looking at a 96GB/s link.
I don't know if it's doable and actually in my own hypothesis I arbitrarily decided a 64gb/s link doable at reasonable cost. We clearly need engineers here
Then for the EDRAM size I assume that you expect tile base rendering (deferred or not) to be the standard. I've neither academic knowledge on that matter but 1MB sounds like both too big and not big enough. To some extend I wonder if modern ROP are not already optimize to sort of "render in tile" /make the most out of the color and Hi/Z buffers, they do way more out of a given amount of bandwidth that the ROP in say the RSX (those in Xenos are as basic as can be). To some extend you lose a lot of the convenience of forward rendering and I wonder if you could not rip the benefit of tile rendering on a standard GPU using appropriate tile size.
Last edited by a moderator:
As for nextgen CPUs, how hard would it be to push very wide SIMDs (like 512 bit), or what if MS did something crazy like push a 6 core Xenon with 256 bit wide SIMDs or 2 x 128 VMX units per core in a manner similar to the split FPU approach in Bulldozer?
IllusionistK
Newcomer
A Pitcairn Pro is essentially the same size as a 90nm Xenos sans the EDRAM. If we include the EDRAM, then this gives you a 64MB pool.
Thus my previous speculation of a Pitcairn Pro, OoO Power7 quad, 64MB of EDRAM.
However, I don't think EDRAM is as beneficial as it once was, and 80mm**2 is a rather large chunk of die.
If MSFT decide to nullify the EDRAM, maybe we'll see a Tahiti Pro with 1792SPs.
Thus my previous speculation of a Pitcairn Pro, OoO Power7 quad, 64MB of EDRAM.
However, I don't think EDRAM is as beneficial as it once was, and 80mm**2 is a rather large chunk of die.
If MSFT decide to nullify the EDRAM, maybe we'll see a Tahiti Pro with 1792SPs.
the reserved cpu L3 was an arbitrary 1MB out of some 80MB total cpu edram, but in fantasy land it can be 160MB so no tile required
and at the moment i'm still trying to understand if tbr is the future or only some exotic elitary tech
french toast
Veteran
Wouldn't 64mb be useless for 1080p? i thought we already discussed a minimum of 128mb preferably 256mb edram..which would be an insane cost.
What kind of memory setup would you use for that set up? we need a minimum of 4gb high bandwidth.
I would be expecting the raw 'flops' number to more or less equal to a high end single gpu when it is released, just like last generation, of course x times the flops does not mean the same overall performance transfered to games..as ALUs are cheaper in hardware than TMUs and ROPs...this is where the performance marketing will focus on.
Saying that, the efficiency and general technology has moved on substantially since the xenos, i wonder if anyone has an example of a 240sp modern gpu to compare against the xenos? i bet all things being equal the newer model would be substantially faster.
So i believe we will see hardware that is 10 times the real gaming performance of x360 minimum, maybe not 10 times the amount of every metric, but overall performance considering enhanced technology and efficiency.
I do believe that microsoft has a very profitable xbox division now, and unlike the xbox it would not be put out to pasture, it would be sold as a budget console, and still make $$ back through live.
This would give them some breathing room, as it is not the console its self that will be counted it will be the xbox division as a whole, and that will be making a profit in the billions of $$ whether or not microsoft sells 720 at a loss or not.
If you take the competiton starting to hot up from Apple,Android, Wii U, onlive and a rejuvinated Sony. for Microsoft to even bother with a new system they will have to make it powerfull enough for consumers to bother with it in 3 years time, when mobile phones will be more powerfull than the 360 in 12 months.
2000+ ALU GPU..maybe vliw5 for bang4buck@ 650-700mhz- custom built.
QUAD POWER PC with 4X SMT, fully OoO with 256bit SIMD.
256bit controller with 6gb GDDR5.
Kinect 2 + HDD lumped in with every box.
Either Blue ray or some propreity flash storage..some kind of game DRM
Run on a modified windows 8.
Blue tooth & NFC.
Would sell at a more premium price point say £350 and at a significant loss, it is expensive but the 360 would hog the low end and make this thing worth investing in over competing platforms.
Side note: Much is made of this, but is it really so compelling that mobile tech may surpass an 7 year old console soon? (and looking at Vita, which looks quite less than 360 so far yet is quite more than any phone or tablet to date, mobile may not even be able to do that anytime soon).
I dont necessarily think so. Next gen consoles will put mobile 8 years behind again (heck, the gulf between say a 7950 and mobile is hard to even comprehend). Non-mobile tech looks to always have a huge advantage, as one would expect.
I would have said : "It's nice to dream."
If you're referring to my earlier post, 128MB was just a hypothetical question. 64MB would be feasible if you didn't use MSAA.
french toast
Veteran
Well i do think that it is compelling..especially if you measure the power outfrom the wall of a 360 in 2013..even including a 28nm shrink and compare it to a state of the art mobile phone, the mobile will likely out perform it in every metric, and do it consuming a fraction of the power.
That is pretty incrdible, if you were to rewind back to 2005 and picture one day you would be able to have something more powerfull in your hand and phone from it you would of thought it was impossible..even after 7/8years.
I don't see that happening next year though, maybe in 3 years.
(That would make it 10 years to reach the same performance in a ridiculous fraction of the power consumption.)
D
Deleted member 13524
Guest
Hum.. looking at ARM SoC roadmaps, I find that near impossible to happen.
Unified Shader performance may eventually be near there, as the shader performance per transistor/per power consumption has been steadily growing.
RAM amount will also be higher, of course has high-end phones are already coming with 1GB.
But for most other specs (memory bandwidth, CPU floating-point performance, TMU+ROP performance and others), the smartphones are still way behind a X360 or a PS3.
For example, the Adreno 200 is said to have a single Vec4+Scalar shader unit (whereas the X360 has 48). With Adreno 320 (probably not standardized until 2013) having 15x the performance of Adreno 200, this means its performance should still be quite a bit lower than half a Xenos.
So while Smartphone SoC performance will undoubtedly surpass the PS360 some day, I doubt that'll happen before 16nm or 11nm, somewhere between 2014 and 2015.
Of course, this is smartphones I'm talking about.
I think it's feasible that in 2013 we'll see x86/ARM chips made for tablets/netbooks with Rogue or AMD GCN GPUs capable surpassing the PS360 in graphics capabilities. I'd expect AMD's top end Samara using DDR4 to easily match Xenos in graphics power. Maybe nVidia Logan too.
Last edited by a moderator:
So...are the console companies supposed to build systems that'll be able to outperform mobile devices ten years from launch? Is that the target now - the impossible march of technological progress, irrespective of the cost?
Besides, that example isn't at all impossible. The Game Gear and Atari Lynx completely trounced consoles and computers from 8 years earlier. It's to be expected. Whatever we can do now in a desktop, we'll be able to do in a handheld in 5-10 years. Not shocking at all.
If the purpose of such comparisons is to identify a performance target for next-gen, then people clearly need to be reminded of what a console is and how the engineers approach the problem. It's a mass-consumer entertainment device that needs to provide an entertainment experience that people will buy, whether that's due to the experience's novelty or quality or whatever. The purpose of the console designers is to put together the best box that'll make the most money. Traditionally this has been by stuffing the release box with as much hardware as possible knowing it'll cost-reduce over the years while still being desirable, making the end result profitable (if you're lucky). If it's seen that a less powerful, less expensive box can be made that still has appeal and will make sales without costing the earth, maybe with an eye on a shorter lifespan, or refreshing the lifespan with extras, or changing to an evolutionary path a la iPod, than that's perfectly acceptable irrespective of how things used to be done. If this thread existed for this gen, who woud have predicted 2x Gamcube for Nintendo's next hardware? And though not the hardware I'd have chosen, it did the job of making them a truckload of money. thus suggestions of low-power GPUs and the like shouldn't be poo-pooed just on account of being less than the best. Within the long-term strategy, it may fit nicely. What if the next Xbox releases at $200 and plays next-gen Halo, COD and Gears etc. in a massive upgrade to the current aged XB360? Are we to believe that a lot of existing long-term 360 owners won't want to upgrade and would rather stick with the current ancient experience because the improvements aren't good enough based on the current tech envelope? The right experience at the right price gets the business.
As such, business talk is pretty useless in this thread. Let's just stick to the hardware discussion.
Not me, I would have been 100% positive Nintendo wouldn't dare do such a thing... And would have been completely wrong
Oh I thought we beat that dead horse already so we were having fun about peripheral ideas... ;pAs such, business talk is pretty useless in this thread. Let's just stick to the hardware discussion.
french toast
Veteran
Well from what we were discussing earlier about Adreno 320, i wouldn't use that as a comparison, perhaps the 420, or a high clocked T658..or top of the food chain Rogue setup.
Certainly in the T658 case, as it suports dx11, a high clocked varient of that with some dedicated vram, a multicore Cortex A-15, 2Gb ram with some duel channel DDR4, maybe 4mb cache with cache coherency..maybe not in every single metric, but i bet you could make better games with that setup...
Don't forget, ignoring the Edram for a moment, the x360 bandwidth is only 22.8gbs or something, exynos 5250 already pulls 12.8gbs..and thats due out in a few months...
Intel could easilly stick a high clocked multi cluster Rogue onto its silvermont @22nm and outdo a 360 by next year if they wish..
My point was that Microsoft has to think big if it wants to introduce a console to last another 7 years...else why bother with the above technology coming out in your hand? we are getting to the point soon where that kind of performance in a small screen is what people will only need and will not really see anything better.
The only reason they would consider a more 'modest' setup is if they want a quick 4 year turn around, and to sell the console for a profit..both cases ala Ninty...
Back to topic, I found this interesting piece earlier about an 'upgradable' 360, i think its a really good idea, say they invest heavilly in the gpu/cpu, equip a Kinect 2 + HDD with every console, they could skimp on the ram to say 2GB to cut costs...and then offer a free 'double ya memory' upgrade to every console in 3 years when the prices come down, or maybe take it a bit further and do the same with a slot in GPU...(although i don't think that is a very feasable before you guys slap me down )
Here is the article..what do you think pie in the sky? or maybe?..
http://www.inentertainment.co.uk/20120129/upgradeable-xbox-720-vs-cloud/
I also think Microsoft would loath to use blue ray if they some how could get around it, the only scenario i think they would is if they added that rumoured DRM rubbish to stop second hand resales..other than that i think it would be both cheaper and more beneficial to use a propreity flash storage, then they could save face, reduce the cost AND reduce the over all size of the console(no royalty payments to Sony..).
Developers would also get to only buy the amount of flash they need tailored to the exact game size..no redundant disk space,Cut down on noise and if its good quality flash streaming shouldn't be problem like it is on blueray currently, the prices of flash are decreasing year on year.
I have already speculated my top end hardware spec, but i have revised it here with more details;
....£350 RRP...Launched holiday 2013.
>Full SOC, Manufactured and designed by IBM&AMD, built @ FOXCONN.
>Custom built 2000 ALU GPU, VLIW 5. @700mhz -improved evergreen class.
>Octo core PPC cpu OoO 4x SMT @3.2ghz 6mb cache- 1 core dedicated for Kinect + system.
>6GB GDDR5 256bit bus unified.
>40GB HDD shipped with every console + Kinect 2 built in.
>WIFI N+Bluetooth+NFC+wireless HDMI.
>All games native 1080p, with 8x MSAA 16x AA.
>Newly designed 360 controller, maybe a small screen and motion control ala Dreamcast+sixaxis.
>Proprietry flash storage with DRM baked in to increase revenue.
>Funky custom built cooling technology..water?
>Faulty chips/poor yields would be cut down and harvested for the redesigned and budget x360 slim.
>Built on 28nm high-k and would Run on custom built windows 8 which will unify the eco system.
With a likely newly designed xbox live+App store, plus the cutting out of second hand games, the xbox division as a whole will be making shed loads of money..this system would be the flagship for the whole microsoft brand together with windows 8, the box would sell at a significant loss at first, but the division as a whole would be making substantial profit, unlike previos years, the move to 20nm would arrive soon after and allow some significant cost cutting.
Certainly in the T658 case, as it suports dx11, a high clocked varient of that with some dedicated vram, a multicore Cortex A-15, 2Gb ram with some duel channel DDR4, maybe 4mb cache with cache coherency..maybe not in every single metric, but i bet you could make better games with that setup...
Don't forget, ignoring the Edram for a moment, the x360 bandwidth is only 22.8gbs or something, exynos 5250 already pulls 12.8gbs..and thats due out in a few months...
Intel could easilly stick a high clocked multi cluster Rogue onto its silvermont @22nm and outdo a 360 by next year if they wish..
My point was that Microsoft has to think big if it wants to introduce a console to last another 7 years...else why bother with the above technology coming out in your hand? we are getting to the point soon where that kind of performance in a small screen is what people will only need and will not really see anything better.
The only reason they would consider a more 'modest' setup is if they want a quick 4 year turn around, and to sell the console for a profit..both cases ala Ninty...
Back to topic, I found this interesting piece earlier about an 'upgradable' 360, i think its a really good idea, say they invest heavilly in the gpu/cpu, equip a Kinect 2 + HDD with every console, they could skimp on the ram to say 2GB to cut costs...and then offer a free 'double ya memory' upgrade to every console in 3 years when the prices come down, or maybe take it a bit further and do the same with a slot in GPU...(although i don't think that is a very feasable before you guys slap me down
)Here is the article..what do you think pie in the sky? or maybe?..
http://www.inentertainment.co.uk/20120129/upgradeable-xbox-720-vs-cloud/
I also think Microsoft would loath to use blue ray if they some how could get around it, the only scenario i think they would is if they added that rumoured DRM rubbish to stop second hand resales..other than that i think it would be both cheaper and more beneficial to use a propreity flash storage, then they could save face, reduce the cost AND reduce the over all size of the console(no royalty payments to Sony..).
Developers would also get to only buy the amount of flash they need tailored to the exact game size..no redundant disk space,Cut down on noise and if its good quality flash streaming shouldn't be problem like it is on blueray currently, the prices of flash are decreasing year on year.
I have already speculated my top end hardware spec, but i have revised it here with more details;
....£350 RRP...Launched holiday 2013.
>Full SOC, Manufactured and designed by IBM&AMD, built @ FOXCONN.
>Custom built 2000 ALU GPU, VLIW 5. @700mhz -improved evergreen class.
>Octo core PPC cpu OoO 4x SMT @3.2ghz 6mb cache- 1 core dedicated for Kinect + system.
>6GB GDDR5 256bit bus unified.
>40GB HDD shipped with every console + Kinect 2 built in.
>WIFI N+Bluetooth+NFC+wireless HDMI.
>All games native 1080p, with 8x MSAA 16x AA.
>Newly designed 360 controller, maybe a small screen and motion control ala Dreamcast+sixaxis.
>Proprietry flash storage with DRM baked in to increase revenue.
>Funky custom built cooling technology..water?
>Faulty chips/poor yields would be cut down and harvested for the redesigned and budget x360 slim.
>Built on 28nm high-k and would Run on custom built windows 8 which will unify the eco system.
With a likely newly designed xbox live+App store, plus the cutting out of second hand games, the xbox division as a whole will be making shed loads of money..this system would be the flagship for the whole microsoft brand together with windows 8, the box would sell at a significant loss at first, but the division as a whole would be making substantial profit, unlike previos years, the move to 20nm would arrive soon after and allow some significant cost cutting.
I guess you know that a AMD A8-385O, so 4 OoO cores and 400sp @600MHz has a TDP of 100W?
GDDR5 memory controller consume more juice. It' already a 225mm2 chip?
Then 6GB of GDDR5 seems complicated on a 256GB bus may be doable but still that's a lot of memory to place on the mobo (12 I think).
I get it that its a "top end design" but I don't think it doable, would be too big too hot.
EDIT
As for the article I believe that Onlive is not a threat to console as the infrastructure is not ready. May be it can take off in some "newly developed country". It like the Onlive concept, I'm subscribing at the time but will stop soon as quality is not good enough, the experience is unstable (most likely because my connection) it goes from good enough for quiet some genre to "deal breaking". I don't how the thing work on a great connection but I suspect that latencies are still too high for many competitive mp games. Overall it can become an establish actor and set its own niche but that's it.
On the other hand I believe that there is room from client-server relationship but I won't go there as there is a proper thread for this
As for the upgradable console, it's a bad idea imho. Introducing new significant peripherals that potentially split your user base is already a iffy move, it's unclear the third party editors support you will receive, the device has to be successful.
GDDR5 memory controller consume more juice. It' already a 225mm2 chip?
Then 6GB of GDDR5 seems complicated on a 256GB bus may be doable but still that's a lot of memory to place on the mobo (12 I think).
I get it that its a "top end design" but I don't think it doable, would be too big too hot.
EDIT
As for the article I believe that Onlive is not a threat to console as the infrastructure is not ready. May be it can take off in some "newly developed country". It like the Onlive concept, I'm subscribing at the time but will stop soon as quality is not good enough, the experience is unstable (most likely because my connection) it goes from good enough for quiet some genre to "deal breaking". I don't how the thing work on a great connection but I suspect that latencies are still too high for many competitive mp games. Overall it can become an establish actor and set its own niche but that's it.
On the other hand I believe that there is room from client-server relationship but I won't go there as there is a proper thread for this
As for the upgradable console, it's a bad idea imho. Introducing new significant peripherals that potentially split your user base is already a iffy move, it's unclear the third party editors support you will receive, the device has to be successful.
Last edited by a moderator:
Do you have a source for that figure? I just read a review of an A8-3850, and the entire system pulled 90 watts at load.
D
Deleted member 13524
Guest
Those 100W aren't measured as if they were thermally relevant (or even battery relevant) consumption values, it's just a fixed ceiling for absolute maximum instant power consumption, making it a bit irrelevant in practical terms.
And it isn't really fair to use a 7 month-old chip which was constantly delayed and already came pretty late by itself.
A more "just" comparison to the state-of-the-art will be in a month or so, when Trinity numbers come out.
- Status
Similar threads
- Locked
