Wii U hardware discussion and investigation *rename

Status
Not open for further replies.
Ok guys...not been on this thread a while...any solid info on the wuu components?..picked up a few pages back that 2gb is confirmed....but rather alarming from my own point view is that a monstrous 1gb is reserved for the OS!!..Xbox 360 used what?..48 MB!!...I know times have changed and all..but seriously 1gb!?

How about the gpu?..still guestimates thinking around hd 4670 kind of performance with some minor changes...
Cpu?..I hope we have some more positive news on this area...last time I was around these parts there was talk of a triple core Wii cpu running at 1.6ghz...please tell me that is seriously wrong...please?? :/
 
Shifty Geezer said:
Closed systems are used in heat-pipes and vapour-chambers. The problem i see with channels to access each layer of a stack is they'd be very small, making it very hard for fluid to flow and evaporate and get out again. I think liquid cooling would have to work on the surface of the chip, with heat conducting through each layer. With the layer so thin, is that really such a problem? Dunno what the thermal conductivity of the silicon is.
Click to expand...

Who says the layers have to be that close together. If the space between layers was several microns, that might be enough for a cooling solution like this. Sure the chip would be taller, but if the rest of the cooling system was done right, that wouldn't make a difference. Additionally, you could stack the layers so that the coolest layer is on the bottom and the hottest on the top. That way the extra heat generated by the hotter layers doesn't need to travel upward through the cooler layers.
 
Additionally, at least in theory, you'd be cooling the chips top AND bottom (instead of just the top, like today). That's twice the area to dissipate heat. Not sure if that's even possible today, though...
 
TheWretched said:
Additionally, at least in theory, you'd be cooling the chips top AND bottom (instead of just the top, like today). That's twice the area to dissipate heat. Not sure if that's even possible today, though...
Click to expand...

Check those link I posted, quite a bit of info on how these things can be cooled. You're not far off!
 
It's not a Wii CPU, but a more current IBM design http://en.wikipedia.org/wiki/PowerPC_400#PowerPC_470
Still rumored to be three cores at 1.6GHz. With maybe a few instructions that I found similar to AMD 3Dnow! added back for 100% compatibility with the Wii CPU (that's speculation from this thread)

I dunno why we know it's 1.6GHz. But even at 1.6GHz it's not entirely terrible as it doesn't suffer the inefficiencies of 3.2GHz Xenon and PPE (a quick and dirty comparison would be AMD Bobcat at 1.6GHz and Pentium 4 at 3.2GHz)
 
Last edited by a moderator:
TheLump's link talks about cooling channels within the chip of about 200 x 150 microns.

To put this into perspective, human blood capillaries can be as small as 5-10 microns in diameter. And blood flow in these vessels is driven predominantly by the gentle peristaltic motion of muscles on the surrounding tissues (and to a a lesser extent the heart).

If blood, which is far far far more viscous than water (or any other cooling fluid used in cooling stacked chips), can be pumped easily through miniscule capillaries much smaller than the channels designed by the chip makers, why not water through said channels?

Ergo, it's not too much of an issue. As you can use a very small low flow high head pump, e.g. a peristaltic device, that can be manufactured cheaply and will run on very low power.

Alternatively, a much much better way is to design a thermosyphon system, that works off the same principles as heat pipes.

In other words almighty is correct. You can basically design an integrated heat-pipe system, using phase-change cooling (far more efficient), capillary action for fluid motion through the small channels in the chip, and natural convective currents to carry the heat away from the chip. It will basically look just like a standard heatpipe on one end, and then the other would be physically attached to the chip packaging.
If you choose the right cooling fluid, and design your pipe to operate so that the fluid boils at a reduced pressure, thus low temperature, you would be able to cool quite a large chip (100W+) with relative ease. Heat pipes alone can transfer upto 1500W of heat (perhaps more), so it should defininately be possible. Plus they are already used in consoles.

The key is in the physical design of the unit, and I believe that the engineers that work at these large electronic technology organisations have been making strides doing just that ;-)
 
TheLump said:
Yeah thats the probelm I weas seeing. Once it gets to such a small level, how efficient can the flow rate be, if it can flow at all?
Click to expand...

It's actually not that much of a problem at all. The smaller the aperture the less work you have to do to force the liquid through the channels, as capillary action will take over and the mass transfer process will become diffusion-driven. It's similar to how a sponge soaks up water after being placed on a flat puddle, or how water can travel up a piece of paper when you dip one end in some water. It's exactly how heatpipes work. This is also beneficial for the intended application as just like with heat pipes there's no mechanical work requirement (i.e. no need for a pump) because the fluid flows on its own.


TheLump said:
Either way, this seems to be the process thats being mooted here

http://www.monolithic3d.com/3d-chip-cooling.html

And here:

http://www.eetimes.com/electronics-news/4077160/IBM-GIT-demo-3D-die-with-microchannel-cooling

Intersteing quote from the first link:

Click to expand...

Great links and great info... thanks for these :)
 
Prophecy2k said:
If blood, which is far far far more viscous than water (or any other cooling fluid used in cooling stacked chips), can be pumped easily through miniscule capillaries much smaller than the channels designed by the chip makers, why not water through said channels?
Click to expand...
I'm not saying it can't happen, but I wonder how well it can work. You're talking such tiny quantities of fluid that the amount of heat they can store is minuscule, requiring a high flow rate to keep the temperature down.

There's going to be some solution, obviously. Probably best for a discussion on processor cooling techniques rather than Wii U speculation, though. ;)
 
Shifty Geezer said:
I'm not saying it can't happen, but I wonder how well it can work. You're talking such tiny quantities of fluid that the amount of heat they can store is minuscule, requiring a high flow rate to keep the temperature down.

There's going to be some solution, obviously. Probably best for a discussion on processor cooling techniques rather than Wii U speculation, though. ;)
Click to expand...

Oh sure, in my mind however the chip would be submerged within the liquid (rather than a series of tube feeding the channels), so the heat flow would not so much be driven by fluid flow through the channels. Rather the channels would provide permeation for the liquid to saturate and nucleate boil within the chip. The system would self-regulate as long as it was well designed and the vapour could be continuously condensed at the cold end of the cooling device. It's no different to heatpipes in essence.

But you're right... this is a conversation for another thread ;-)
 
Prophecy2k said:
It's actually not that much of a problem at all. The smaller the aperture the less work you have to do to force the liquid through the channels, as capillary action will take over and the mass transfer process will become diffusion-driven. It's similar to how a sponge soaks up water after being placed on a flat puddle, or how water can travel up a piece of paper when you dip one end in some water. It's exactly how heatpipes work. This is also beneficial for the intended application as just like with heat pipes there's no mechanical work requirement (i.e. no need for a pump) because the fluid flows on its own.
Click to expand...

I see, thanks. I was referring more to what Shifty later mroe eloquently said, about the amount of heat that can be transferred by such a small amount of liquid; but as you say this may not be as much of an issue as it sounds. And indeed, I suppose the bightest minds in America wouldn't be persuing this process if there weren't some sound science behind it, so I'll trust in that ;)

Prophecy2k said:
Great links and great info... thanks for these :)
Click to expand...

Pleasure. Its really a rather interesting topic but it's complexity also highlights how unlikely it is this has anything to do with WiiU - so at least the conversation was productive in that sense!
 
Shifty Geezer said:
I'm not saying it can't happen, but I wonder how well it can work. You're talking such tiny quantities of fluid that the amount of heat they can store is minuscule, requiring a high flow rate to keep the temperature down.
Click to expand...
Convection is a lot faster than conduction, like orders of magnitude faster. So even if you don't have a lot of fluid, and even if it doesn't flow super-fast, it's still moving thermal energy a lot faster than it dissipates through a conductive media, even one with much higher conductivity than copper. Your car's engine block has a relatively small amount of fluid circulating through it compared to the volume of the metal it has to cool.

Further, convection enhances conduction by increasing the temperature differentials in the material itself. By cooling the surface of the object, it allows heat from the core to dissipate to the surface more quickly, where it is removed and cooled by the fluid again.

You only need a flow rate high enough to remove heat significantly faster than it's being conducted into the fluid by the material, which is really not all that much. For example, In turbine blades, we use convective cooling by pumping air through pinhole-sized apertures in the blade. It creates a very, very tiny boundary layer of "cool" air (by "cool," I mean "not nearly as hot as burning jet fuel") around the blade, but that small amount of flow on its own is enough to prevent the blades from melting.
 
Last edited by a moderator:
Tell Microsoft or Sony to drop me a next generation machine off and I'll strap these bad boys on and make it squeal..

IMG_0880.jpg


-40c should do nicely

:cool:
 
3M and IBM have been working on an adhesive to cool 3D stacks, specifically designed for high-power 3D chips. It's a few micron thick and conducts heat vertically between silicon layers, still thin enough to allow TSV. No idea how close they are to a commercial product though.

Maybe a sheet of vertical carbon nanotubes would be perfect for this too... it's not too far fetched as they already make lateral heat speader with carbon nanotech today. (graphene?)

http://www-03.ibm.com/press/us/en/pressrelease/35358.wss
 
fearsomepirate said:
Convection is a lot faster than conduction, like orders of magnitude faster. So even if you don't have a lot of fluid, and even if it doesn't flow super-fast, it's still moving thermal energy a lot faster than it dissipates through a conductive media, even one with much higher conductivity than copper. Your car's engine block has a relatively small amount of fluid circulating through it compared to the volume of the metal it has to cool.

Further, convection enhances conduction by increasing the temperature differentials in the material itself. By cooling the surface of the object, it allows heat from the core to dissipate to the surface more quickly, where it is removed and cooled by the fluid again.

You only need a flow rate high enough to remove heat significantly faster than it's being conducted into the fluid by the material, which is really not all that much. For example, In turbine blades, we use convective cooling by pumping air through pinhole-sized apertures in the blade. It creates a very, very tiny boundary layer of "cool" air (by "cool," I mean "not nearly as hot as burning jet fuel") around the blade, but that small amount of flow on its own is enough to prevent the blades from melting.
Click to expand...

Now THAT is pretty cool :D
 
Status
Not open for further replies.

Similar threads

Rys
Reopening Architecture and Products + rules and guidelines changes
3 4 5
Replies
90
Views
13K
Shifty Geezer
Shifty Geezer
Back
Top