By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - Gaming - XBONE having problems with eSRAM yield. Crazy Buttocks confirms.

Slimebeast on 07 June 2013
walsufnir said:
Slimebeast said:
walsufnir said:
Slimebeast said:
This is hilarious.

If it's true like Ethomaz said, that the Xbone APU is more expensive than the PS4 APU and will remain so over the whole generation, and yet the PS4 is 50% more powerful, now that must be seen as a serious engineering failure.

I'm so pissed off. How can they fail so bad?

I want xbone to be able to give PS4 a match, a good fight. Not become laughing stock.

 

It's not the apu, it is the whole chip with APU, memory controller, move units and esram. Of course this is more expensive.

All right, the whole chip.

But it's the same problem - how can something be "of course more expensive" and at the same time weaker than the competition?

What's the benefit of this design decision? That the machine will be able to communicate faster with separate OPs (which makes it fast to switch between different programs, like gaming, Kinect, Skype, TV etc)? That's the only argument I've heard so far.


You are looking at it the wrong way.

$ony and MS started developing their "next-gen" years before, with a goal to achieve, also technically. MS thought it would be a good way to switch to x86 but had good experience with edram in 360 - given that they also have experience in 3d (they build directx) they got specs from amd and together a solution was found to increase bandwidth between apu and ddr3 -> esram. To make it even better (in their view) also Move-units were developped to act as fixed-function-units.

Overall this all adds up - move-units, memory-controller, apu, memory-controller - in one package. It is not "weaker" as you see it. It just integrates more stuff in one package.

What we see here is two different companies with two different ways of implementing a "new gen". After all is said and done it's easy to say "wow, that's crap".

But that's what I feared.

That at the end of the day it will turn out there are no benefits to the Xbone's design over PS4's.

That MS engineers simply made less good choices than the competition, back in the day when the groundworks of next gen console design was made.

That's what I feared and that's what pains me.

If I still was an Xbox fan (and I still am to a degree) I would be banging my head to a wall right now. This would haunt me through the whole coming generation.



Around the Network
darkknightkryta on 07 June 2013
Pemalite said:
darkknightkryta said:

I know very well what "flops" means.  You're also glossing over the part where it takes more processing power to do non integer calculations.  I mean, most geometry transformations are in floating point numbers.


Actually, most of the processing at the start of the graphics rendering pipeline is floating point heavy, not just geometry transformations. - The end of the pipeline is mostly integer heavy, integer math is generally faster, game developers relied more heavily on it towards the end of this generation to extract more performance.

As for my prior point, seems it flew completely over your head.

The fact that FLOPS are used a a standard way to determine a processor's power?  Please tell me how that went over MY head.  FLOPS are used cause it's very difficult to store and calculate a floating point number.  And yes, integer processing is faster, it's also less accurate which is why you'd want do leave the numbers as is if possible.



Nsanity on 07 June 2013

Greenberg and Major Nelson are saying that this isn't true at all.



Legend11 on 07 June 2013

Microsoft should sue NeoGAF/CBOAT if this isn't true, that might be the only way to get people to believe them.



walsufnir on 07 June 2013
Slimebeast said:
walsufnir said:


You are looking at it the wrong way.

$ony and MS started developing their "next-gen" years before, with a goal to achieve, also technically. MS thought it would be a good way to switch to x86 but had good experience with edram in 360 - given that they also have experience in 3d (they build directx) they got specs from amd and together a solution was found to increase bandwidth between apu and ddr3 -> esram. To make it even better (in their view) also Move-units were developped to act as fixed-function-units.

Overall this all adds up - move-units, memory-controller, apu, memory-controller - in one package. It is not "weaker" as you see it. It just integrates more stuff in one package.

What we see here is two different companies with two different ways of implementing a "new gen". After all is said and done it's easy to say "wow, that's crap".

But that's what I feared.

That at the end of the day it will turn out there are no benefits to the Xbone's design over PS4's.

That MS engineers simply made less good choices than the competition, back in the day when the groundworks of next gen console design was made.

That's what I feared and that's what pains me.

If I still was an Xbox fan (and I still am to a degree) I would be banging my head to a wall right now. This would haunt me through the whole coming generation.


Ehm, no, you still don't understand. It's not only MS engineers. This is the outcome of a team made of MS engineers, amd engineers, memory engineers and probably guys who work for the factory building the chip for the Xbox One. What you are confusing is that this single chip is the whole system which is (of course) wrong and especially also true for the design. What you will see in the end onscreen is not only depended on the apu.



Around the Network
slowmo on 07 June 2013
Legend11 said:
Microsoft should sue NeoGAF/CBOAT if this isn't true, that might be the only way to get people to believe them.


If it isn't true then they're likely already being tracked down by Mircosoft's lawyers.  It's not like Microsoft would come out and say we are going to sue them, they'd just quietly find out who they are then slap down a huge suit on them.  If as they say it's not true then these people may regret thinking they were safe posting this on a website.  I guess time will tell



Slimebeast on 07 June 2013
walsufnir said:
Slimebeast said:
walsufnir said:
 


You are looking at it the wrong way.

$ony and MS started developing their "next-gen" years before, with a goal to achieve, also technically. MS thought it would be a good way to switch to x86 but had good experience with edram in 360 - given that they also have experience in 3d (they build directx) they got specs from amd and together a solution was found to increase bandwidth between apu and ddr3 -> esram. To make it even better (in their view) also Move-units were developped to act as fixed-function-units.

Overall this all adds up - move-units, memory-controller, apu, memory-controller - in one package. It is not "weaker" as you see it. It just integrates more stuff in one package.

What we see here is two different companies with two different ways of implementing a "new gen". After all is said and done it's easy to say "wow, that's crap".

But that's what I feared.

That at the end of the day it will turn out there are no benefits to the Xbone's design over PS4's.

That MS engineers simply made less good choices than the competition, back in the day when the groundworks of next gen console design was made.

That's what I feared and that's what pains me.

If I still was an Xbox fan (and I still am to a degree) I would be banging my head to a wall right now. This would haunt me through the whole coming generation.


Ehm, no, you still don't understand. It's not only MS engineers. This is the outcome of a team made of MS engineers, amd engineers, memory engineers and probably guys who work for the factory building the chip for the Xbox One. What you are confusing is that this single chip is the whole system which is (of course) wrong and especially also true for the design. What you will see in the end onscreen is not only depended on the apu.

My assumption is based on the claim by ethomaz that:

"But I can say for sure... the Xbone's APU is more complex to produce end and at least 40% more expensive than the PS4's APU.

MS will never catch PS4 in APU costs in this generation... the difference is simple:

PS4: small (3 billion transistors) and simple (CPU, GPU and memory controller)
Xbone: big (5 billion transistors) and complex (CPU, GPU, memory controller, eSRAM, DataMove, etc)

The percentage of the GOOD chips per Wafer for PS4 will be always better than Xbone too."

If it's true that the Xbone APU is significantly more expensive than the PS4's, despite having a much weaker GPU, then surely that also must reflect into the cost of the whole system. In simple terms, their design choice gives less bang for your buck from a gaming perspective.

On the other hand, PS4 has significantly more expensive RAM, but that also brings tangible value.

It's still a fact that having a large chip brings with it additional costs due to inefficacy, at least early in a gen when the manufacturing process is less mature and gives smaller yields.



walsufnir on 07 June 2013
Slimebeast said:
walsufnir said:
Slimebeast said:
walsufnir said:
 


You are looking at it the wrong way.

$ony and MS started developing their "next-gen" years before, with a goal to achieve, also technically. MS thought it would be a good way to switch to x86 but had good experience with edram in 360 - given that they also have experience in 3d (they build directx) they got specs from amd and together a solution was found to increase bandwidth between apu and ddr3 -> esram. To make it even better (in their view) also Move-units were developped to act as fixed-function-units.

Overall this all adds up - move-units, memory-controller, apu, memory-controller - in one package. It is not "weaker" as you see it. It just integrates more stuff in one package.

What we see here is two different companies with two different ways of implementing a "new gen". After all is said and done it's easy to say "wow, that's crap".

But that's what I feared.

That at the end of the day it will turn out there are no benefits to the Xbone's design over PS4's.

That MS engineers simply made less good choices than the competition, back in the day when the groundworks of next gen console design was made.

That's what I feared and that's what pains me.

If I still was an Xbox fan (and I still am to a degree) I would be banging my head to a wall right now. This would haunt me through the whole coming generation.


Ehm, no, you still don't understand. It's not only MS engineers. This is the outcome of a team made of MS engineers, amd engineers, memory engineers and probably guys who work for the factory building the chip for the Xbox One. What you are confusing is that this single chip is the whole system which is (of course) wrong and especially also true for the design. What you will see in the end onscreen is not only depended on the apu.

My assumption is based on the claim by ethomaz that:

"But I can say for sure... the Xbone's APU is more complex to produce end and at least 40% more expensive than the PS4's APU.

MS will never catch PS4 in APU costs in this generation... the difference is simple:

PS4: small (3 billion transistors) and simple (CPU, GPU and memory controller)
Xbone: big (5 billion transistors) and complex (CPU, GPU, memory controller, eSRAM, DataMove, etc)

The percentage of the GOOD chips per Wafer for PS4 will be always better than Xbone too."

If it's true that the Xbone APU is significantly more expensive than the PS4's, despite having a much weaker GPU, then surely that also must reflect into the cost of the whole system. In simple terms, their design choice gives less bang for your buck from a gaming perspective.

On the other hand, PS4 has significantly more expensive RAM, but that also brings tangible value.

It's still a fact that having a large chip brings with it additional costs due to inefficacy, at least early in a gen when the manufacturing process is less mature and gives smaller yields.

 

Basing on an assumption of ethomaz... is... haha, no, I like him, too, but the numbers are gambled, really.

Costs highly depend on contracts and stuff, we will never really know the truth. Also the high failure rate has been clear to everyone who was involved in the project, that's for sure. If additional costs have to be paid by the customer is also a decision made by the console-manufacturer.



hinch on 07 June 2013
Nsanity said:
Greenberg and Major Nelson are saying that this isn't true at all.

Well of course they would. They work as PR for Microsoft.

And fyi, this has already been debunked (as pointed out by ethomaz). The guy has a history of leaking stuff way ahead of everyone else. He's legit.



ethomaz on 07 June 2013

walsufnir said:

Basing on an assumption of ethomaz... is... haha, no, I like him, too, but the numbers are gambled, really.

Costs highly depend on contracts and stuff, we will never really know the truth. Also the high failure rate has been clear to everyone who was involved in the project, that's for sure. If additional costs have to be paid by the customer is also a decision made by the console-manufacturer.

Even if 100% of the ships are GOOD the Xbone's APU is more expensive than PS4's APU... you can produce more chips for PS4 in a wafer than Xbone... it is 5 billion transistors vs 3 billion transistors.

There are no magical process that can change that... big chips are more expensives than small one... no matter what contract you do or anything else.

The Xbone's APU is 40% bigger than PS4's APU in transistors... MS can make a more dense chip (small transistors and more close to each other) but there a limit for that... at the end the PS4's APU will be small and cheaper.

My prediction...

PS4's APU: 250mm²
Xbone's APU: 410mm²