Predict: The Next Generation Console Tech
- Thread starter Acert93
- Start date
- Status
I was hoping for fall 2014 with exotic technology
How about fall 2013 with the biggest and best that proven technology can offer.
"One in the tush is better than two in the bush." TWSS.
Bagel seed
Veteran
Hmm. Countdown to E3 means no sneak peak or preview before then like OurColony or other viral campaigns as they did for the 360.
Smells like delayed launch to 2014
Smells like delayed launch to 2014
How about fall 2013 with the biggest and best that proven technology can offer.
Except it wont contain the biggest tech. The best tech it more than likely but a cut down version of it for certain.
I don't think waiting until 2014 for exotic tech is any better though since by 2014 its not going to be particularly exotic anymore compared with 2014 PC gpus.
Hmm. Countdown to E3 means no sneak peak or preview before then like OurColony or other viral campaigns as they did for the 360.
Smells like delayed launch to 2014
Giving Nintendo 2 years headstart doesnt seem like a good idea.
You mean the same way that 7870 / 7950 is a cutdown version of 7970? Not really, it's not going to be off the shelf.
7850 / 7870 class is more realistic, - in relation to whatever is the highest end hardware of the time of course and with some level of customisation, e.g. fewer rops, different memory structure etc...
Bottom line though is not to expect 79xx / 89xx level grunt.
Last edited by a moderator:
Definitely not in anyway, never mind just on paper
Yep that's my point, so it wont be the biggest and best technology on offer at the time, it'll be an upper midrange equivilent of the best technology at the time (at best in all likelyhood).
The biggest thing that is holding nextgen consoles from reaching their full potential with 2013 cpu/gpu architectures is sadly cooling.
I would really like for MS or Sony to create console that is aimed to be cooled by air only after first chip revision. Launch console will therefore have to be cooled by small and efficient liquid cooling rig [20-40 bucks], with one fan that will cool not only that rig but provide nice airflow over internal structure.
Of course this wont happen.
I would really like for MS or Sony to create console that is aimed to be cooled by air only after first chip revision. Launch console will therefore have to be cooled by small and efficient liquid cooling rig [20-40 bucks], with one fan that will cool not only that rig but provide nice airflow over internal structure.
Of course this wont happen.
The biggest thing that is holding nextgen consoles from reaching their full potential with 2013 cpu/gpu architectures is sadly cooling.
I would really like for MS or Sony to create console that is aimed to be cooled by air only after first chip revision. Launch console will therefore have to be cooled by small and efficient liquid cooling rig [20-40 bucks], with one fan that will cool not only that rig but provide nice airflow over internal structure.
Of course this wont happen.
The Surface Pro hardware has some innovative cooling solutions. I did expect to see them expanded to the Durango.
MS of 2012, when it comes to hardware, is much, much, more talented and experienced than in 2005.
Blazkowicz
Legend
From what I think to know the cooling fluid is transported to a big heatsink, with big slow fan(s) blowing on it. It's effective (when your metric is amount of watts removed) but from a cost perspective if you need a heatsink and fan anyway why not do air-cooling and do away with the pipes and stuff.
If you refer to the small contained, and somewhat cheap "mini-watercooling" products they sure look nice but I've read bad things about at least one of them (ended up noisier than a regular cooler)
I know of one consumer product which shipped with standard liquid cooling : PowerMac G5 (with up to two hungry CPUs). There were leaks.
The easiest would be for a hungry console to be physically big. Maybe something stand-up, as was the PC-FX?, or just shaped like an amplifier/receiver.
If you refer to the small contained, and somewhat cheap "mini-watercooling" products they sure look nice but I've read bad things about at least one of them (ended up noisier than a regular cooler)
I know of one consumer product which shipped with standard liquid cooling : PowerMac G5 (with up to two hungry CPUs). There were leaks.
The easiest would be for a hungry console to be physically big. Maybe something stand-up, as was the PC-FX?, or just shaped like an amplifier/receiver.
Recently "negative air pressure" liquids were introduced into watercooling rigs.
http://news.softpedia.com/news/Negative-Pressure-Liquid-Cooling-Prototyped-294944.shtml
yeah yeah definitely only on paper, many reasons.
succces of durango/orbis an all this "next gen" will be very important for AAA oriented developers. Moreso than graphical parity (not "framebuffer parity") of pc version( Pc marketd drifted away from AAA in comparison to last transition ...).I bet we will see many situations like just cause, GRAW, bad company etc when pc version was lacking significant feauters/was late outrageous port. Basicly, in many cases i expect soft touched last gen versions for pc, not for technology sake, but marketing/small target base on pc side
Exlusives will be out of reach for budget reasons no matter how many fans we screw in pc case.And i feel there might more of them, sonys and ms studios are growing quickly... and in this day and age will be main selling points
In addition to that, last transition devs had to learn how to deal with all these mulicore/mulithreded paradigm and even shaders in case od sony developers. This time it will be easier to bring out the performance quickly.
So yes there will be cards with paper spec much higher, especially when maxwell arrives year after... Just like last time , in many cases underperforming(x1000/7800 in graw/jc, few performing better(if game is pc port need for speed etc.)Soon after that we'll hear about new API needed (of course on another round of cards), which should bring "revolution" in pc side , but in practice will be used to make up for inefficiencies and allow second and third generation of durango/orbis games to come to pc and that 2014 cards will fade out.
This vision is somehow grim for tiny bit of pc graphics enthusiast left in me. Sadly, considering all of this, and remembering how this gen unfolded (eg. after seven years we don't have anything remotely on a scale of FEAR on pc which came out 4 years after last consoles), IQ gap will shrink again, an multi platforms are mostly uprezzed console ports.
For me this is like being happy, that in 2006 i can play quake 3 in my native monitor resolution....
If these situation is going to repeat i'll gladly see those fancy new cards only in b3d discussions.
The problem with that is it isn't actually +75% peak FLOPS since they didn't get the number right for the HD 7870 (which is 2.56 TFlops instead of 2.25 TFlops, reducing that figure to +50%). Which begs the question why should I trust this random table for the future gpus if they can't even get the figures for the current ones correct? Oh yes and the percentage increase for all of sp, dp and texture filtering are all _different_ (and none of the 3 numbers is correct...)
That said such a tflop increase would be doable in the projected area / transistor count (adding 8 CUs would add about 40mm²), though I'm sceptical - imho any die size / transistor count increase beyond adding 4 CUs should be spent elsewhere, and I also can't quite see how (together with the clock increase) you could still fit this into the exact same TDP.
All the AMD talk made me want to make some silicon budget comparisons.
CPU
AMD Zactate (Bobcat / E-350) was 75mm^2 on 40nm in Winter 2011 with 2 CPU cores (1.6GHz) and 80 stream processors (500MHz) with 18W TDP. (Bulldozer, just to compare, has 8 CPU INT cores and was 315mm^2 on 32nm.) AMD’s new stream architecture has 64 stream processors per CU.
http://www.anandtech.com/show/4134/the-brazos-review-amds-e350-supplants-ion-for-miniitx
But it gets interesting as the GPU is nearly 75% of the Bobcat core:
http://www.chip-architect.com/news/AMD_Ontario_Bobcat_vs_Intel_Pineview_Atom.jpg
A CPU is more than the core, so there needs to also be room for various cache sizes and levels, memory controllers, and so forth. Moving forward Jaguar is threatening to introduce quite a few enhancements to Bobcat so it is likely the core size will inflate quite a bit, especially as caches are very important to performance. 28nm is theoretically 50% smaller than 40nm yet Jaguar cores will, again, have a bit of enhancements and likely require more local memory so it is not likely to fit 4 cores into the same area as Bobcat cores, even with the die shrink.
But even being generous and saying 2 Bobcat cores + cache, memory controller were half a die at 75mm^2 (so about 38mm^2 for 2 cores) it would seem 2 Jaguars could (conjecture) fit into that die area on 28nm and 4 Jaguar cores into 75mm^2. 150mm^2 would be a rough guestimate to the total area needed for 8 Jaguar cores.
Because Al is so awesome: http://beyond3d.com/showthread.php?t=62651
Xenon was 167mm^2 on 90nm in Fall 2005.
Cell was 235mm^2 on 90nm in Fall 2005.
>> 4 Jaguar cores are conjecturally going to be about half, or less, the silicon budget of 2005 consoles and much lower TDP.
>> 8 Jaguar cores would be roughly in the ball part of the silicon budget of the 2005 consoles but with a similar or lower TDP.
GPU
Pitcairn (Radeon 78570) was 212mm^2 on 28nm in Winter 2012 with 20 CUs (1280 stream processors), 80 TMUs, and 32 ROPs with various models clocking in from 860MHz to 1000MHz with various units disabled on a 256bit bus with a total GPU board TDP ranging from 130-175W.
Xenos was 262mm^2 (182mm^2 GPU and 80mm^2 Daughter Die) in Fall 2005.
RSX was 258mm^2 in Fall 2005.
Various things to consider is 28nm maturing as well (density, performance, TDP), various tradeoffs to maximize the primary limiting factor (TDP) between fewer units/higher clocks and more units/lower clocks, and that high frequency GDDR5 uses a lot of power—not necessarily just the chips/# but the memory controller. It is hard to guess the power needed for eDRAM (especially if it is a cache instead of a write buffer like Xenos) and the size, and important if stacked memory and a Silicon interposer may be used.
>> Pitcairn is in the general size class, if not on the slightly small size, compared to consoles in 2005 (or about the same if 32-64MB of eDRAM was introduced).
Looking at these numbers, and the SA snapshot of a chip with memory on a Silicon Interposer, it seems really likely one could do the following:
~ 325mm^2 SoC
>> ~ 75mm^2 : 4 Jaguar cores
>> ~ 200mm^2 Pitcairn class GPU with some CUs disables
Put two memory modules (stacked?) on the same SI and you lower the power and ramp up the bandwidth. You could even go with a relatively large bus from the entire SI to a more general memory pool.
So in theory you could have full “HAS” with a higher end GPU and CPU on the same die and a boatload of lower power memory next door. The only real major drawback seems to be the large step back in total CPU cores and peak throughput—but I don’t know how reasonable 8 CPU cores would be going with a SoC.
Anyways, it seems to my armchair (bad?) math and guesses the AMD rumblings of a Pitcairn class + Jaguar cores SoC, possibly on an SI, seem very reasonable based on last gen BOM budgets for the silicon.
I am sure a lot of people would consider such a design very elegant and balanced.
CPU
AMD Zactate (Bobcat / E-350) was 75mm^2 on 40nm in Winter 2011 with 2 CPU cores (1.6GHz) and 80 stream processors (500MHz) with 18W TDP. (Bulldozer, just to compare, has 8 CPU INT cores and was 315mm^2 on 32nm.) AMD’s new stream architecture has 64 stream processors per CU.
http://www.anandtech.com/show/4134/the-brazos-review-amds-e350-supplants-ion-for-miniitx
But it gets interesting as the GPU is nearly 75% of the Bobcat core:
http://www.chip-architect.com/news/AMD_Ontario_Bobcat_vs_Intel_Pineview_Atom.jpg
A CPU is more than the core, so there needs to also be room for various cache sizes and levels, memory controllers, and so forth. Moving forward Jaguar is threatening to introduce quite a few enhancements to Bobcat so it is likely the core size will inflate quite a bit, especially as caches are very important to performance. 28nm is theoretically 50% smaller than 40nm yet Jaguar cores will, again, have a bit of enhancements and likely require more local memory so it is not likely to fit 4 cores into the same area as Bobcat cores, even with the die shrink.
But even being generous and saying 2 Bobcat cores + cache, memory controller were half a die at 75mm^2 (so about 38mm^2 for 2 cores) it would seem 2 Jaguars could (conjecture) fit into that die area on 28nm and 4 Jaguar cores into 75mm^2. 150mm^2 would be a rough guestimate to the total area needed for 8 Jaguar cores.
Because Al is so awesome: http://beyond3d.com/showthread.php?t=62651
Xenon was 167mm^2 on 90nm in Fall 2005.
Cell was 235mm^2 on 90nm in Fall 2005.
>> 4 Jaguar cores are conjecturally going to be about half, or less, the silicon budget of 2005 consoles and much lower TDP.
>> 8 Jaguar cores would be roughly in the ball part of the silicon budget of the 2005 consoles but with a similar or lower TDP.
GPU
Pitcairn (Radeon 78570) was 212mm^2 on 28nm in Winter 2012 with 20 CUs (1280 stream processors), 80 TMUs, and 32 ROPs with various models clocking in from 860MHz to 1000MHz with various units disabled on a 256bit bus with a total GPU board TDP ranging from 130-175W.
Xenos was 262mm^2 (182mm^2 GPU and 80mm^2 Daughter Die) in Fall 2005.
RSX was 258mm^2 in Fall 2005.
Various things to consider is 28nm maturing as well (density, performance, TDP), various tradeoffs to maximize the primary limiting factor (TDP) between fewer units/higher clocks and more units/lower clocks, and that high frequency GDDR5 uses a lot of power—not necessarily just the chips/# but the memory controller. It is hard to guess the power needed for eDRAM (especially if it is a cache instead of a write buffer like Xenos) and the size, and important if stacked memory and a Silicon interposer may be used.
>> Pitcairn is in the general size class, if not on the slightly small size, compared to consoles in 2005 (or about the same if 32-64MB of eDRAM was introduced).
Looking at these numbers, and the SA snapshot of a chip with memory on a Silicon Interposer, it seems really likely one could do the following:
~ 325mm^2 SoC
>> ~ 75mm^2 : 4 Jaguar cores
>> ~ 200mm^2 Pitcairn class GPU with some CUs disables
Put two memory modules (stacked?) on the same SI and you lower the power and ramp up the bandwidth. You could even go with a relatively large bus from the entire SI to a more general memory pool.
So in theory you could have full “HAS” with a higher end GPU and CPU on the same die and a boatload of lower power memory next door. The only real major drawback seems to be the large step back in total CPU cores and peak throughput—but I don’t know how reasonable 8 CPU cores would be going with a SoC.
Anyways, it seems to my armchair (bad?) math and guesses the AMD rumblings of a Pitcairn class + Jaguar cores SoC, possibly on an SI, seem very reasonable based on last gen BOM budgets for the silicon.
I am sure a lot of people would consider such a design very elegant and balanced.
COPS N RAPPERS
Regular
when you read behind the words, It seems microsoft could really be thinking about injecting their console with plenty of power. they wouldn't be bringing that up if it wasn't foremost in their minds.
though I expect MS to be very quiet on the specs this time, last gen and the one before everyone was pretty open, now there's closed off campuses and...... timers. still, If the IBM rumors are true you can expect a very special complimentary gpu to follow. i hope for a custom job.
- Status
Similar threads
- Locked
