Xbox One (Durango) Technical hardware investigation

[GrimThorne]

I would say that you can't say that unless you know what criteria they were designing for. I think cost, manufacturability, power consumption, etc may have been more important than just raw performance.

Look I understand the need for energy efficiency and cost effective manufacturing, but most issues, especially those related to fabrication/maximizing yields and assembly usually take care of themselves over time. Microsoft is a hardware manufacturer, this isn't foreign territory to them.

No matter, we'll find out just what was prioritized and what wasn't soon enough. And I know I keep going over this, but it's still just puzzling to me. The depth of the talent on this team could have have made their OWN chip (not what I'm suggesting) for Microsoft.

It's like hiring ninjas and then instructing them to make paper snowflakes.

 

[GrimThorne]

You're comparing Durango to (desired) performance, assuming the intention of the engineers is Moare Powarrr. What if they've got the performance of a 7790 in 80% of the cost and 75% of the power consumption, coupled with a 50% in RAM costs by going with DDR3? Then their goals, which weren't low hanging, were achieved by good engineering.

The only way to determine the effectiveness of the engineers is to know what targets they were designing for, not what we want them to design. Same with Wii U - the engineers managed to get comparable performance to PS360 in a much smaller and cheaper package. We can grumble all we like about performance, but that's not due to the ineptitude of the engineers; that's a fault of the decision makers giving them low-power targets.

All I'm saying is that if Durango turns out to be what's been revealed, I'm just a little disappointed. But you're right, whatever they did was at Microsoft's behest.

 

[Shifty Geezer]

All I'm saying is that if Durango turns out to be what's been revealed, I'm just a little disappointed. But you're right, whatever they did was at Microsoft's behest.

There's a whole other thread for how people feel about MS's new console. This one is about discovering and understanding what hardware is inside that box.

 

[GrimThorne]

There's a whole other thread for how people feel about MS's new console. This one is about discovering and understanding what hardware is inside that box.

There is actually one lingering curiosity that I have about the Durango specs. On the memory schematics, has anyone properly identified or explained the vacant space between the move engines and the GPU?

 

[kots]

A big thing being overlooked is that the new Xbox need not be more powerful than what we already know. Hardware specs alone don't make games, developers do.

If you feel that way , that Durango is good enough as it is , why make a dozen posts about overclocking ?

And what it all ultimately boils down to is that it's quite simple enough to address certain performance related concerns by lowering the resolution. 1680x1050. I say get use to it or some form of it now, as we'll probably be seeing plenty of it on Durango.

Why , because MS says so ? Really ? And "get used" to it ? No thanks. Seems that there are other alternatives so i won't have to accept compromises.

 

[KSterson]

Clocking 200 Mhz higher is no small adjustement. It's probably pushing the whole tdp to +40w something. It may even recquire some thermal design change, or with the power supply (everything you try to fit really close to hardware specifics). A small adjustement in clock is still possible but i don't think 200 Mhz is likely at this point. My guess 800 Mhz is not a random clock if both appeared to have set their gpu at this clock, it's certainly the best ratio in perf/consumption/therm/.. in a SoC (they are usually cloked lower with way smaller gpu parts in them). Like 500 Mhz was the only rational frequency on X360/PS3.

 

[function]

Why , because MS says so ? Really ? And "get used" to it ? No thanks. Seems that there are other alternatives so i won't have to accept compromises.

All devices are interwoven chains of compromises. You can't avoid them.

 

[SenjutsuSage]

If you feel that way , that Durango is good enough as it is , why make a dozen posts about overclocking ?

As I said, it's good enough as is. And it isn't technically overclocking, because that range of GPU is designed, at its core, to handle higher clocks. My belief that they possess the capability to pursue higher clocks on that range of GPU without too much trouble has absolutely nothing to do with the fact that I feel, quite confidently, about the specs as we currently know them.

This isn't about my desires, so much as it's about what I think Microsoft can achieve rather easily with the range of GPU they've chosen. I agree that they don't have to build the most powerful system that they can, but the least they can do is get an extra

Why , because MS says so ? Really ? And "get used" to it ? No thanks. Seems that there are other alternatives so i won't have to accept compromises.

Call it whatever you like, and feel free to explore whatever alternatives you prefer. That's not the point. Developers are simply best served designing to get the most out of the hardware. If that means 1680x1050 instead of 1080p, I don't see what the big deal is. I don't see what the fuss is about with 1080p, and never have. An incredible looking hi-def game will still be an incredible looking hi-def game, whether it's in 1680x1050 or 1080p.

I'm not expecting a full 200MHZ clock speed increase to 1GHZ. I expect a minimum clock speed of 900-925MHZ for the Durango GPU, and If it's to be lower, then I expect at least 850-875MHZ, but 800MHZ just doesn't sound right to me for the range of GPU that we're looking at.

Only way 800MHZ makes sense to me is if Durango's GPU turns out to be at least a 16 Compute Unit part, as opposed to what's currently rumored. Unless they plan on putting this hardware in one of the tiniest possible frames or yields are just that problematic, then it seems that a clock higher than 800MHZ is easily within their ability to achieve without altering their plans too much.

 

[GrimThorne]

Looking at Durango's specs, I would say that overclocking would end up being a bigger mess than it's worth. I mean if that additional 200Mhz was worth that much to Microsoft they could have gone with a Pitcairn and gotten the same results without the heat and instability.

I don't think Microsoft would risk it. Would they even have time to run the necessary stress tests?

 

[AlphaWolf]

Looking at Durango's specs, I would say that overclocking would end up being a bigger mess than it's worth. I mean if that additional 200Mhz was worth that much to Microsoft they could have gone with a Pitcairn and gotten the same results without the heat and instability.

I don't think Microsoft would risk it. Would they even have time to run the necessary stress tests?

When they start mass producing their product and find that tolerances allow for higher clock speeds, they simply run them at a higher clock speed. They can choose a value that fits their envelope.

A larger part would also require more power and produce more heat. 16CU pitcairn (radeon 7850) at 860 MHz uses substantially more power (around 20W) than a 10CU part (radeon 7770) at 1GHz.

Choosing a high clock doesn't necessarily equate to more heat and instability over choosing a larger part. AMD already has a number of products running GCN architecture over 1GHz.

 

[bagofsuck]

i am confused..the GPU is a 7770? or a 7790? :S

 

[kots]

Looking at Durango's specs, I would say that overclocking would end up being a bigger mess than it's worth. I mean if that additional 200Mhz was worth that much to Microsoft they could have gone with a Pitcairn and gotten the same results without the heat and instability.

I agree ...why go overclocking when you can get a better gpu in the first place ?

I don't think Microsoft would risk it. Would they even have time to run the necessary stress tests?

It's not about risk ... to raise the Mhz you need a beefier cooling solution - and that costs money . Seeing the innards of Durango , it's pretty obvious they are a bit stingy about that.

 

[DJ12]

i am confused..the GPU is a 7770? or a 7790? :S

Neither, it's based on one of them, which one depends who you want to believe.

Those that think the leaks are right or those that don't believe them.

Personally, there has been too much noise from fans of the xbox at every suggestion of mysterious-ness that I don't believe any of it now.

I'm guessing it's 7770 based as the leaks say. There's no LOGICAL reason why MS would suddenly change their plan because another console looks more powerful on paper as they must've expected that from the start with the route they took.

 

[SenjutsuSage]

When they start mass producing their product and find that tolerances allow for higher clock speeds, they simply run them at a higher clock speed. They can choose a value that fits their envelope.

A larger part would also require more power and produce more heat. 16CU pitcairn (radeon 7850) at 860 MHz uses substantially more power (around 20W) than a 10CU part (radeon 7770) at 1GHz.

Choosing a high clock doesn't necessarily equate to more heat and instability over choosing a larger part. AMD already has a number of products running GCN architecture over 1GHz.

This is more or less what I'm thinking also. I feel that Microsoft -- I doubt by coincidence -- find themselves in a position where they can more than likely benefit from higher GPU clock speeds without using too much power in the process, which is why it wouldn't surprise me if they decided to take advantage of it in the end if they find that their yields are good enough.

And just to add, this isn't something that I think Microsoft would be doing in a reactionary manner. I think they are firmly set in their plans regardless of what anybody else is doing, and they likely always intended on taking advantage of higher clocks for their GPU if circumstances allowed for it, and I think they intentionally created the circumstances that they knew would allow for it. To those that keep calling it overclocking, it honestly isn't, not for the range of GPU that Microsoft has chosen it isn't.

Greater than 1GHZ could be considered overclocking. 900-925MHZ or even 850-875MHZ would still be quite a bit below the default speeds for this range of GCN GPU, hence not overclocking in the least.

 

[Scott_Arm]

Neither, it's based on one of them, which one depends who you want to believe.

Those that think the leaks are right or those that don't believe them.

Personally, there has been too much noise from fans of the xbox at every suggestion of mysterious-ness that I don't believe any of it now.

I'm guessing it's 7770 based as the leaks say. There's no LOGICAL reason why MS would suddenly change their plan because another console looks more powerful on paper as they must've expected that from the start with the route they took.

I'm not saying this rumour is true, and I think it's not likely to be true, but why would this rumour suggest MS "suddenly changed their plan"? Beta is beta, not final. It's entirely possible that they could have been intending for something higher than 800 MHz from the start. If the final clock speed is a floating target within a range, then rolling out a dev kit with the minimum makes sense until you can finalize it. It could be that they're iterating through a number of cooling solutions for the final box design, or that they're working on yields as they increase production of the unit, hoping for something above 800 MHz. And of course it's also very possible that it will just be 800 MHz as it is in the beta kits.

 

[DuckThor Evil]

A larger part would also require more power and produce more heat. 16CU pitcairn (radeon 7850) at 860 MHz uses substantially more power (around 20W) than a 10CU part (radeon 7770) at 1GHz.

Choosing a high clock doesn't necessarily equate to more heat and instability over choosing a larger part. AMD already has a number of products running GCN architecture over 1GHz.

7850 still uses the large pitcairn 20CU chip and probably consumes more than a native 16CU would. It also has double the memory chips at a slightly higher clocks. The performance difference is quite big and 7850 has better performance per watt, at least according to techpowerup. (except at 1024*768)

http://www.techpowerup.com/reviews/AMD/HD_7850_HD_7870/27.html

 

[ERP]

I'm not saying this rumour is true, and I think it's not likely to be true, but why would this rumour suggest MS "suddenly changed their plan"? Beta is beta, not final. It's entirely possible that they could have been intending for something higher than 800 MHz from the start. If the final clock speed is a floating target within a range, then rolling out a dev kit with the minimum makes sense until you can finalize it. It could be that they're iterating through a number of cooling solutions for the final box design, or that they're working on yields as they increase production of the unit, hoping for something above 800 MHz. And of course it's also very possible that it will just be 800 MHz as it is in the beta kits.

At some level it doesn't matter what was in the Beta kits, what matters is what they tell devs will be in the final boxes.
It is common for early Beta kits to use early silicon steppings with bugs that result in reduced performance, you certainly inform devs of that and your documentation reflects the expectation of the final hardware.
Final hardware is usually not a surprise to devs when they get it.

 

[SenjutsuSage]

I'm not saying this rumour is true, and I think it's not likely to be true, but why would this rumour suggest MS "suddenly changed their plan"? Beta is beta, not final. It's entirely possible that they could have been intending for something higher than 800 MHz from the start. If the final clock speed is a floating target within a range, then rolling out a dev kit with the minimum makes sense until you can finalize it. It could be that they're iterating through a number of cooling solutions for the final box design, or that they're working on yields as they increase production of the unit, hoping for something above 800 MHz. And of course it's also very possible that it will just be 800 MHz as it is in the beta kits.

More or less. I think we make a huge mistake when we assume that Microsoft are suddenly changing their plans or operating in a reactionary manner. This could have been in the cards all along, just like 8GB of GDDR5 was likely in the cards for Sony for quite a bit longer than we all think. I don't believe they suddenly made some ultra last minute change to their plans. it was likely something they were always thinking about, but it wasn't something they were prepared to commit to doing until it actually became possible for them to do so. I see more or less a similar situation playing out on the other side.

If MS doesn't do it, it will ultimately be because it just didn't fit with their plans. I'm not saying this based on rumors, I'm saying that I think this is what they will do based on what the clock speeds and TDP on other GCN GPUs in this power range are. I'll be very surprised if the final clock speed on the Durango GPU isn't somewhere between 875-925MHZ

 

[DJ12]

I'm not saying this rumour is true, and I think it's not likely to be true, but why would this rumour suggest MS "suddenly changed their plan"? Beta is beta, not final. It's entirely possible that they could have been intending for something higher than 800 MHz from the start. If the final clock speed is a floating target within a range, then rolling out a dev kit with the minimum makes sense until you can finalize it.

The leaks aren't based on what's in the dev kit, but what MS have told developers they will be offering.

 

[SenjutsuSage]

I guess, if anything, what surprises me is why does this seem so unrealistic or impossible for some when Sony damn near surprised just about everybody with 8GB of GDDR5? You would think that of the possible differences from the leaks, a higher clock speed on the GPU would be seen as far less unbelievable than 8GB of GDDR5.

Even digitalfoundry suggested it was impossible that Sony could add in 8GB of GDDR5, and these folks had people confirm to them the validity of the specs, or at least what had been communicated to developers.