Xbox One (Durango) Technical hardware investigation

[3dilettante]

Question..

Is it right to assume that more RAM memory, also increases heat?
So going from 8 to 12, would this increase heat as well?

Because the clock bump on the GPU was so "small", could it be that MS also chose to increase other areas as well, such as CPU speed and RAM?

Just tossing this question out there...

There is a very minor effect for power consumption based on RAM capacity. The dominant factor is RAM speed and voltage, although there may be more complex relationships between this factors in the DRAM manufacturer's binning process.

Attempts to measure a difference in PCs that try to control for speed and voltage haven't found anything noticeable, at least not considering sources of measurement variation that dwarf the difference.

 

[COPS N RAPPERS]

This is good news, Microsoft is getting some payback on their "wealthy" cooling system, it's a minor speed bump, but also a minor bump on heat/watt. The heat will most likely be kept in check by the cooler, just a few more RPM's. And the power usage is paid by us..

Good news!

It's great news that Microsoft actually attempted a GPU increase. though the risk for a clock jump is pretty risky. The heat sink and fan i think did play a minor roll in it but they now have to work a little harder; which may raise the noise level.

I was hoping for a TDP increase and a licensed GPU revision higher. The development costs would slightly go up but they would save on the testing time involved and maximize on the flop per clock increase.

 

[Tap In]

posted on GAF Albert Penello MS

Originally Posted by Albert Penello
Hey guys

Lots of interesting comments on the GPU upgrade.

Let me put this out there. I know there is doubt about the “truth” of what we say, so some of you will believe me, and others won’t. But here goes…

We set aggressive targets for reliability, performance, yields, and noise. Those things always have to be balanced. We want this box to have rock-solid reliability. We want it to be DEAD quiet (and let me tell you, X1 is quieter than the new Xbox 360 we just released). And we wanted killer game performance. But those targets are in conflict with each other.

What we’ve found through the development process is we were able to actually exceed our goals on the thermals and acoustics. This gave us headroom to increase the clock speed without any hit to noise, reliability, or heat, so we took the opportunity to bump the GPU. I get it’s only 6% or so, but that could translate to a few FPS in the real world.

I know there are many conspiracy theories out there about how and why we make decisions. I can tell you – this was something we were hoping to be able to do for a while. So we were prepared for this. Nobody should worry this puts us at any risk or people are scrambling at this decision.

 

[3dilettante]

It's great news that Microsoft actually attempted a GPU increase. though the risk for a clock jump is pretty risky.

It's more of a clock tweak. The fact that it was done this late in the game should indicate that the risk is very low to null.

The heat sink and fan i think did play a minor roll in it but they now have to work a little harder; which may raise the noise level.

Would it work harder if the chip can support the higher clock while still meeting the power target?
This is a good possibility.

I was hoping for a TDP increase and a licensed GPU revision higher. The development costs would slightly go up but they would save on the testing time involved and maximize on the flop per clock increase.

There's about three months to go, which is just a little more than a smidgen too late for a spec change and a redesign of the GPU silicon.

 

[pjbliverpool]

posted on GAF Albert Penello MS

A good honest answer. I like it.

 

[expletive]

A good honest answer. I like it.

Me too but the real lesson here is that you (or just MS?) can barely do anything these days without pissing off half the internet.

 

[XpiderMX]

I know that E3 is a noisy place but, do we have some info about noise from XOne?

 

[COPS N RAPPERS]

posted on GAF Albert Penello MS

Very interesting, so they still want to keep it to the same goals.

It's more of a clock tweak. The fact that it was done this late in the game should indicate that the risk is very low to null.

yeah, because they picked a safe zone; very little testing was necessary. The risk factor on a 53 mhz increase should be minimal.

Would it work harder if the chip can support the higher clock while still meeting the power target?
This is a good possibility.

the TDP on an increase could be a minor jump, though i'm thinking it will stay the same.

There's about three months to go, which is just a little more than a smidgen too late for a spec change and a redesign of the GPU silicon.

Yeah, they've had since the reveal to do something like that. As of last month now, time is of the essence...unless they want low shipping figures for maximizing their time. I still don't know what the shipping figures will be.

 

[3dilettante]

I know that E3 is a noisy place but, do we have some info about noise from XOne?

That would require having the final systems running with final fan control software, final coolers, and final enclosures at E3.

For the Xbox One, the rumors have it that all of these elements weren't available until very recently.

 

[Ekim]

There is probably still more to come... Let's wait and see.

 

[Rangers]

A good honest answer. I like it.

I dont. They weren't willing to stretch their extremely aggressive silence/heat goals for a little bit more performance. Only delivering an upclock if it didn't budge thermals/noise *at all*. Doesn't sit well with me. They likely left performance on the table. Maybe they could have easily got to say, 950, a meaningful increase, if they were willing to accept just a little more noise and heat.

I had forget that with mandatory installs all around, these systems wont be spinning the optical drives too much. They should be really quiet.

BTW, with an upclock now reality, we can say B3D vindicated, other forums shamed

LATER EDIT Penello addresses my concerns later

Gaffer

What bothers me is that decibel figures take precedence over performance.

Penello's reply

That's an interesting point, let me say two things.

First - I wouldn't jump directly to that conclusion. I said these things had to be in balance. There are other ways to get good acoustic performance.

I'm not saying you don't make tradeoffs against those things, but it wouldn't be correct in assuming we made noise a priority.

Second (and you corrected yourself) people use the box for a lot of media functions. I think it would take a beating if we were even close to the noise level of the 2005 360.

As I remind people sometimes. We have a console that is roughly 8x - 10x the performance of last gen, depending on how you define it. It's in a case that is only ~ 10% larger than the launch 360. And yet it's quieter than the 3rd major revision we did 7 years in.

I'm not sure what he meant, could be typical all things to all people damage control, but it makes me feel slightly better.

"There are other ways to get good acoustic performance". Maybe they beefed up the heatsink in order to do the upclock?

Also, "wouldn't be correct in assuming we made noise a priority" is unfortunate wording I'm sure. Of course it's SOME priority, I'm sure he meant more like "not the only priority".

And I disagree with his second point. I'd certainly be willing to accept a box say, "only", lets say, 25% quieter than 2005 Xbox for more performance. You would still be significantly quieter than launch last gen launch.

And the way I see it, later revisions can always focus more on the quiet aspect. Just as this gen.

 

[XpiderMX]

Stupid question: Can the frequencies level (higher and lower) be altered by a firmware update?

 

[3dilettante]

I dont. They weren't willing to stretch their extremely aggressive silence/heat goals for a little bit more performance. Only delivering an upclock if it didn't budge thermals/noise *at all*. Doesn't sit well with me. They likely left performance on the table. Maybe they could have easily got to say, 950, a meaningful increase, if they were willing to accept just a little more noise and heat.

I had forget that with mandatory installs all around, these systems wont be spinning the optical drives too much. They should be really quiet.

The more you bump things up, the more the ignorable factors start to matter.
Keeping everything iso-power and close to the original configuration is nice because it probably means they didn't have to leave the bounds of their yield targets and the guard-bands on their validation and reliability testing.

20% is more likely to force an uptick in voltage and require a review of the system around the APU. There's a much greater chance that they need to reevaluate the design if clocks, volts, cooler, and yield targets get shifted. We don't know what reliability tests they did and for what parameters. Redoing that is non-trivial.
All this a few months out is asking for trouble.

 

[RobertR1]

This is good news, Microsoft is getting some payback on their "wealthy" cooling system, it's a minor speed bump, but also a minor bump on heat/watt. The heat will most likely be kept in check by the cooler, just a few more RPM's. And the power usage is paid by us..

Good news!

I question the mindset of someone so poor that they are worried about the energy price jump from the minor spec bump yet buying a $500 consoles + games and accessories.

Do you know anyone in that situation?

 

[Rangers]

We might even get some more out of Penello as he's still posting. In response to one asking for confirmation of 1.31 TF

I'm confirming with the silicon team how this impacts ESRAM and TFLOPS. I'm pretty sure but I don't like to post without being 100% sure.

Unfortunately, I can't comment on any rumors. I try to come on and provide clarity and more detail around things we've already announced. But I cannot for obvious reasons start confirming or denying rumors.

And I think the second part may have even been referring to the 12GB RAM rumors.

However if he confirms the flops then he just confirmed the number of shaders, and inferiority in a major spec, dont know if MS wants to go there, so I'm not holding my breath on that.

Maybe I'm crazy, but the way Penello acts 12GB seems possible. He also ignores questions about it on twitter.

It just seems if it wasn't true, he would behave differently.

It would be terribly easy to say "The RAM is still 8GB as we announced".

 

[Scott_Arm]

The more you bump things up, the more the ignorable factors start to matter.
Keeping everything iso-power and close to the original configuration is nice because it probably means they didn't have to leave the bounds of their yield targets and the guard-bands on their validation and reliability testing.

20% is more likely to force an uptick in voltage and require a review of the system around the APU. There's a much greater chance that they need to reevaluate the design if clocks, volts, cooler, and yield targets get shifted. We don't know what reliability tests they did and for what parameters. Redoing that is non-trivial.
All this a few months out is asking for trouble.

They also have to worry about power in idle because of all those regulations in countries that don't like components that are "turned off" sapping power from the grid. That's another consideration if you get into voltage bumps, rather than just looking at the top end spec.

 

[Hardknock]

What would they even use an extra 4 GBs on? More things could be multitasked but anything else? Seems like more of a marketing bullet point than a significant benefit in the real world.

 

[Scott_Arm]

What would they even use an extra 4 GBs on? More things could be multitasked but anything else? Seems like more of a marketing bullet point than a significant benefit in the real world.

There is a limit on how much data the GPU and CPU can touch per frame, but more memory means more information for varied game objects. My only question is how feasible it would be to actually fill that memory from what's likely a 5400 rpm HDD.

 

[(((interference)))]

Hmm, so it was upclocked after all - though it's even less than the 75 mhz rumoured.
Being only a measly 53 mhz, as said before, they must be getting it for free, with negligible impact on anything else (yields, thermals, longevity etc).

If for example, the supposed 10% GPU OS reserve is true, well, you just negated most of that loss.

That's not a rumour, it's pretty much fact:
http://kotaku.com/the-five-possible-states-of-xbox-one-games-are-strangel-509597078

1) Running: The game is loaded in memory and is fully running. The game has full access to the reserved system resources, which are six CPU cores, 90 percent of GPU processing power, and 5 GB of memory. The game is rendering full-screen and the user can interact with it.

 

[astrograd]

Hmm, so it was upclocked after all - though it's even less than the 75 mhz rumoured.
Being only a measly 53 mhz, as said before, they must be getting it for free, with negligible impact on anything else (yields, thermals, longevity etc).



That's not a rumour, it's pretty much fact:
http://kotaku.com/the-five-possible-states-of-xbox-one-games-are-strangel-509597078

The part about 90% seems officially confirmed now since I think Multerer stated as much in a recent OXM article on the OS and why it's good to keep the partitions separate.

Btw, Panello is over at GAF and he basically said they did the upclock because it didn't have any effect whatsoever on their thermals or acoustics so they went with it.