Predict: The Next Generation Console Tech

Status
Not open for further replies.
But it could have many advantages over time. Motherboard could be simpler, internal bandwidth between components huge. Would allow slim versions over a shorter period of time. Expensive to start with but if you can get shrinks in a timely manner total lifetime production costs could be lower.

Single big chip plus memory may be a good lifetime proposition.
 
Low yields mean low performance? sorry, english is not my main language :oops:
Typically low yields means the reverse. The higher performance the part, being larger and running faster, the lower the number of valid chips per press. However, there's really not enough info to determine the cause, even if this rumours is true which we don't know.
 
But it could have many advantages over time. Motherboard could be simpler, internal bandwidth between components huge. Would allow slim versions over a shorter period of time. Expensive to start with but if you can get shrinks in a timely manner total lifetime production costs could be lower.

Single big chip plus memory may be a good lifetime proposition.
Is it not possible to start with discrete components and then create a suitable SOC or integrated chip when cost effective to do so later down the line? I don't think there's any advantage to starting stupidly large and expensive. Large and expensive is hard enough to launch at.
 
Is it not possible to start with discrete components and then create a suitable SOC or integrated chip when cost effective to do so later down the line? I don't think there's any advantage to starting stupidly large and expensive. Large and expensive is hard enough to launch at.

But can HSA/FSA be designed on discrete components? :?:
 
Typically low yields means the reverse. The higher performance the part, being larger and running faster, the lower the number of valid chips per press. However, there's really not enough info to determine the cause, even if this rumours is true which we don't know.

It depends on what compromises have to be made to get yields up. Lowered clocks, disabled functionality, etc.. but I agree that low yielding parts are usually higher performance.
 
Also, if they are making chip with GPU and CPU on one die, plus ESRAM, that could add to complexity which could in return result in low yields. Although, from reading this, it seems like yields are absolutely horrible, so not sure that SOC complexity would be accurate to say.
 
Is it not possible to start with discrete components and then create a suitable SOC or integrated chip when cost effective to do so later down the line? I don't think there's any advantage to starting stupidly large and expensive. Large and expensive is hard enough to launch at.

If you had something like a fast and large shared memory between GPU/CPU that required a fast bus (250+ GB/sec), I think that would be hard, power hungry, and very expensive to do with discrete parts on a PCB. I think it could be done on an interposer layer, but that could be the source of poor yields as well.
 
Is it not possible to start with discrete components and then create a suitable SOC or integrated chip when cost effective to do so later down the line? I don't think there's any advantage to starting stupidly large and expensive. Large and expensive is hard enough to launch at.

If they start with separate chips then they will lose the performance advantages of having the chips combined into one.
 
Gamestop says that next year we will see only one new console, so probably this fall wiiu, then xbox, and in 2014 ps4
 
Gamestop says that next year we will see only one new console, so probably this fall wiiu, then xbox, and in 2014 ps4

Well it could be the other way around Some rumor a while ago stated that Sony were confident that they would make it to the market before MSFT.

Rumors anyway... as for Charlie.D his new piece sounds likeke a cover up.
 
Well... Sony also said (to that rumor of them being definately first), that they only release when they're ready... so it's going both ways^^
 
If you had something like a fast and large shared memory between GPU/CPU that required a fast bus (250+ GB/sec), I think that would be hard, power hungry, and very expensive to do with discrete parts on a PCB. I think it could be done on an interposer layer, but that could be the source of poor yields as well.

Doubtful though that they'd be shopping around if they were already targeting an interposer, as it would be a new process, they'd want to stick with one fabricator as long as possible. Extremely poor yields though could bring a 2.5d solution into play though as one of the main risks would be equalized.
 
Well... Sony also said (to that rumor of them being definately first), that they only release when they're ready... so it's going both ways^^

That was a pretty careful non-answer that doesn't imply anything, let alone counter an independent rumor.
 
Which would render the "Pro" variant useless for most people, as developers/publishers will focus on the low end model, as everyone has that.
Also the contrary can be true because if Kinect needs some extra processing because of the burden it adds to the system then.... It's just a fact more hardware muscle can do that.

Kinect games weren't high watermark as far ax Xbox 360 visuals go.

But it is fair to assume that Xbox 720 won't be a weaker system because of Kinect 2. On the contrary, it is going to be more beefy than it would be without it to process the additional features.
 
Gamestop says that next year we will see only one new console, so probably this fall wiiu, then xbox, and in 2014 ps4

Way more likely PS4 2013 and XB720 2014 imho. Sony has to worry about Japan and losing market share to Nintendo there. Maybe Sony won't release in North America and Europe in 2013, but Japan seems likely to me. I don't think it is just about Sony and Nintendo, Microsoft needs to buffer against any Apple move and 2014 would help with that.

Just my opinion here, but If Microsoft is going with Nvidia Project Denver (64bit ARM CPU), 8 Gigabytes of Hybrid Memory Cube ram, and Holographic disc, then 2014 seems like the sweetspot. Also Unreal Engine 4 will be primed to melt eyeballs. Then in 2015 launch Kinect 2.0 with improved cameras and such and have a Halo Game ready.
 
If one of these two companies delays release to 2014, it will probably be for reasons that was stated in yesterdays SemiAccurate rumor. As of now specs are locked, and there is no way one of them will delay it so they can get things like 3D stacking in console since it probably wouldn't even be possible in 2014 for mass production. I don't think GS guy knows anything at all...
 
BG said Durango dev kit chips (as it said in papers) where without ESRAM, but it will be in retail machines. Charlie said first batch of chips will be produced by the end of February and ready for developers in March. Interestingly, thats the exact same time rumors about dev kits coming in developers hands started and same time BG gave info. Its possible that Charlie had legit info this time around...
 
Seems like the Charlie Rumor is start report by some media...but why it become processor problem lol?
 
Last edited by a moderator:
Status
Not open for further replies.
Back
Top