News & Rumors: Xbox One (codename Durango)

Status
Not open for further replies.
Clocks are subject to change. If MS can get another 100-200 MHz from the APU, I don't see why Sony can't, at which point the argument of chasing higher clocks seems to fall down.

Yes. But imo they have less incentive to do so. It's MS that if rumors are true is much more in need of a performance boost. Considering it will likely come at some cost-yield, heat, etc, one needs to be motivated to do so. Sony might say in the given example, "well we still have comfortably more flops, plus our RAM is faster, no worries". The motivation is lesser.

Also there is less and less time here, even for clock boosts. Sony event was in February, but MS event will be in May. That's 3 months difference. I imagine that will be very near the point where EVERYTHING has to be locked down and finalized. For consoles to be in stores in Nov, production cant be that far away now. It's also occurred to me, I have no idea if they think this way at all, but if they truly did a upclock, it might well behoove them to not mention it on the 21st, to prevent a response before it's too late. They dont have to give a flop number out on the 21st at all. They could list other specs, call it a custom AMD GPU and not give any details, and show games. Sure some devs might know, but it'd likely take a while longer for info to filter that route back to Sony.

Anyways I still think an update to 12Gb is also pretty possible. I consider them about equal in possibility actually. Certainly late RAM upgrades are now developing a pretty solid history in consoles.

The problem is they'd be wasting a ton of money to barely move the needle. Why torpedo your production capacity just to make your system's paper deficit slightly smaller? It's foolhardy to waste potentially billions on the manufacturing of useless chips so that the Durango is only 400 GFlops behind PS4 instead of 600.

We dont know they'd be "torpedoing production capacity", and 10-20% more performance is pretty important, arguably especially for Durango.

How often is "The PS4 has 50% more flops!!" thrown around?? You're telling me if that was cut to 25% it wouldn't be a big deal? LOL.

1.8/1.4=1.285 (~28% faster). The message board wars would be changed quite a bit on that alone.

if we're being realistic, MS could get another 50 MHz and Sony choose not to stress their cooling solution by matching.

You really think that's all? The only thing I have to go on is the 10-12 CU SI parts, and those stock at 1ghz with models even factory overclocked to 1100 mhz readily available at little premium. I understand it's hardly a perfect analogue.

From an armchair perspective 960 mhz, 20%, would seem doable. But I understand there's so much we dont know about. I also understand both being clocked at 800 initially by rumor, probably means it's a sweet spot at the least.

But I do think this idea that these parts in Durango/PS4 can magically go to exactly 800 mhz and not one mhz farther literally cant be true.
 
Clocks are subject to change. If MS can get another 100-200 MHz from the APU, I don't see why Sony can't, at which point the argument of chasing higher clocks seems to fall down.

:???: bbot was just talking about MS upping their clocks to get competitive advantage with the rumoured PS4 specs. Ignoring the whole performance comparison debate, given reported similarities in architecture I don't see how MS could have the option to up the clocks a little but Sony do not, so that avenue seems a dead-end to me. If MS decide to up clocks to compete, Sony should be able to do the same thus undoing the performance advantage. The end result would be two platforms that run a little faster than they would otherwise be running, but also running hotter and possibly with a higher failure rate, and neither console company gaining from the move.

Given that any significant increase in clocks has to serviced by a cooling solution robust enough to handle the increase and quality control testing to ensure reliable operation, its probably much easier to increase clocks earlier in the design/pre production stage versus later.

I dont really believe that MS is going to make any significant changes to durango but making changes in march and april is more feasible than making changes in june and july for a fall product launch.
 
Last edited by a moderator:
Anyways I still think an update to 12Gb is also pretty possible. I consider them about equal in possibility actually. Certainly late RAM upgrades are now developing a pretty solid history in consoles.

I sincerely doubt MS can just add 4GB as if it was nothing or as if it would not change the perchance of accessible memory.
With a higher memory amount and with this memory set up the current bandwidth could really be inadequate.
 
Last edited by a moderator:
http://arstechnica.com/gaming/2013/05/microsoft-next-xbox-will-work-even-when-your-internet-doesnt/

New Xbox will not require an Internet connection to watch Blu Ray, play single-player games ... watch live tv? Does this imply there is a tuner option, either analog or digital?

The 360 can watch live TV now as its one of only 2 MCE (Media Center Extenders) currently available that I'm aware of. The 4GB Slim makes for a damn good, inexpensive, extender as well. So, no, I wouldn't say it necessarily implies a tuner option since there's a very high chance the Durango will be an extender as well. Could very well also support a tuner or tuner pass-through of some sort as well, of course.
 
Since 28nm chips have been in mass production for well over 2 years, would yields really be that bad? It's not as if Microsoft and Sony are trying to build monstrous APUs the size of a GTX 680 right when 28nm was brand spanking new. Heck, the console manufacturers are the most conservative of all the consumer electronics companies when it comes to fabbing larger chips (CPUs, GPUs, memory, and smartphones are all far faster to going to a new process node).

What am I missing here?
 
But they may still block used games?


Well, given the statements made by J. Paul Raines last month

http://www.ign.com/articles/2013/04/03/gamestop-calls-next-xbox-hot-compelling

"We’ve been spending a lot of time with Microsoft,” Raines admitted, “but we have to let them take the lead on [the new console], but it will be a very hot, very compelling device. They are doing some really cool stuff, and I’m eager to hear them start their announcements because I think the world is going to stand up and take notice.”


And, what he said last february after some bad news about Durango blocking used games, where Gamestop shares went down

http://finance.yahoo.com/blogs/the-exchange/gamestop-shares-fall-224923306.html

"our customers want used-game option"

I do think, that no, Ms won´t block used games
 
Microsoft you presume would
Have more room for pushing up clocks, due to the rumoured gpu specs containing substantially less execution resources.

If Microsoft main competitor can also eek out another 150mhz clock bump durango will be in serious trouble.

Personally I don't think we will see anything different, too late and too expensive to start pissing around a couple of months before production.

As a side note, didn't-partially accurate - say durango was getting built at a IBM plant early Last year or something??

Come on the 21st..end the torment already.
 
Since 28nm chips have been in mass production for well over 2 years, would yields really be that bad? It's not as if Microsoft and Sony are trying to build monstrous APUs the size of a GTX 680 right when 28nm was brand spanking new. Heck, the console manufacturers are the most conservative of all the consumer electronics companies when it comes to fabbing larger chips (CPUs, GPUs, memory, and smartphones are all far faster to going to a new process node).

What am I missing here?

That 28nm chips haven't been in mass production for well over 2 years.
 
I sincerely doubt MS can just add 4GB as if it was nothing or as if it would not change the perchance of accessible memory.
With a higher memory amount and with this memory set up the current bandwidth could really be inadequate.

Wouldn't more memory improve the overall performance/experience even with no increase in bandwidth? Especially if the OS footprint is as huge as is rumoured, leaving less available to games than the opposition.

Storing more things in memory is better than waiting for them the load every time. 8GB seems like plenty, but more is always better right?
 
Wouldn't more memory improve the overall performance/experience even with no increase in bandwidth? Especially if the OS footprint is as huge as is rumoured, leaving less available to games than the opposition.

Storing more things in memory is better than waiting for them the load every time. 8GB seems like plenty, but more is always better right?

I am not sure.
ESRAM bandwidth is not shared with the CPU which means that, with 4GB more, the CPU would have to read/write 9GB of RAM at 69GB/s bandwidth (most likely even less) instead than 5GB.
To me it looks like a very unbalanced solution because it would mean more slower memory to deal with.

Why have SRAM is memory amount is not a concern?
Why have 5GB for games if you can have 8GB no problem?

Correct me if I am wrong of course.
 
Last edited by a moderator:
Cost? More of the same RAM wouldn't slow anything down. It'd just give you more RAM, and the DDR3 is orders of magnitude quicker than the BD-ROM or HDD.
 
Could someone with some real in depth knowledge in this area answer this..

How difficult would it be to =
-up clocks of underwhelming cpu to 2ghz.
-up clocks of underwhelming gpu to 1ghz.
-increase ram up to 12gb.
-increase ram speed to a more respectable amount?

All of that or even some of that I don't expect to see the light of day, but would it be a very difficult technological challenge for the engineers to undertake? Would that push up heat by a substantial amount? That doesn't seem too far and above what's possible in theory on discrete pc parts... However this being a custom SOC design I'm guessing it's not directly comparable to what's going on a pc.
 
Cost? More of the same RAM wouldn't slow anything down. It'd just give you more RAM, and the DDR3 is orders of magnitude quicker than the BD-ROM or HDD.

Costs are always a factor but I would leave them aside for now.

Now more RAM would not affect bandwidth but I am not sure it would not affect the percentage of accessible memory as well.
 
Since 28nm chips have been in mass production for well over 2 years, would yields really be that bad? It's not as if Microsoft and Sony are trying to build monstrous APUs the size of a GTX 680 right when 28nm was brand spanking new. Heck, the console manufacturers are the most conservative of all the consumer electronics companies when it comes to fabbing larger chips (CPUs, GPUs, memory, and smartphones are all far faster to going to a new process node).

What am I missing here?

Sony and Microsoft seems to be the most aggressive and not the most conservative outside of Nvidia and AMD gpus especially during product launches. While smartphones processor vendors get to be the most conservative of all.

Sony and MS employs practically brand new designs on top of the newest processes. Thats not as applicable this gen versus previous gens, but Durango and Orbis are custom designs that carry more cpus and CUs than AMD has ever offered. Has AMD released it's 28nm based parts into the wild yet?

Intel employs its tick tock strategy employing new designs on more mature processes then employing mature designs on newer processes. AMD does this also but not always. When it comes to gpu AMD and Nvidia will readily employ new designs on the latest process but only if they have to do so. AMD cpus also benefits from the fact that AMD typically employs their gpus on new processes first before migrating their cpus over.

AMD and Nvidia can afford to launch large gpus on the latest manufacturing process because they are low volume high profit chips. They aren't purchasing huge wafer volumes to accommodate their highend gpus and the profit per viable die is large enough to accommodate poor yields.

Apple will regularly manufactures its biggest chips on older processes. Look at the ipad, the A5X launched at 45nm just over a year ago and the A6X launched last Oct at 32nm. ARM processors are so small that they remain cheap even on older processes which is helped by the fact they serve products with relatively huge profit margins. They also benefit from the fact that their chips are small so even with poor yields will still provide a significant amount of viable dies. Samsung migrated the Exynos 5 chip from 32 to 28 nm for a Q2 2013 launch of the Galaxy 4 and they seem to one of the most aggressive in terms of migrating to newer processes but they still employ an intel like tick tock strategy.
 
Last edited by a moderator:
Status
Not open for further replies.
Back
Top