Article: AMD's Mobile Strategy
By: David Hess (davidwhess.delete@this.gmail.com), January 11, 2012 10:46 am
Room: Moderated Discussions
Megol (golem960@gmail.com) on 1/11/12 wrote:
---------------------------
>Mark Roulo (nothanks@xxx.com) on 1/10/12 wrote:
>---------------------------
>>Jason Chan (jchan@gmail.com) on 1/10/12 wrote:
>>---------------------------
>>>Bill Henkel on 12/29/11 wrote:
>>>---------------------------
>>>> For desktops and workstations, a differentiator could be
>>>> higher single thread performance rather than more cores
>>>> per socket (e.g. a 6 GHz dual-core water-cooled processor).
>>>
>>>The power consumption of a 6 GHz dual-core x86 processor would not need to be higher
>>>than Sandy Bridge-EP so water cooling would not be necessary. Suppose ...
>>
>>There are probably a lot of issues with this, but one obvious one is heat/power
>>*DENSITY*. Trying to pull 150 watts out of a 30 mm^2 die is probably quite a challenge.
>>
>>I think that the PowerPC 970 had this problem ...
>
>Shouldn't a vapor chamber system with the silicon die exposed (boiling on the silicon surface) work?
Heat pipes have a limit on the amount of power they can absorb in a given surface area of the evaporator. Above that limit, film boiling takes place and cooling fails because only gas is in contact with the evaporator. Going from memory, the best designs are limited to 30 watts per square centimeter while 10 watts is more typical.
---------------------------
>Mark Roulo (nothanks@xxx.com) on 1/10/12 wrote:
>---------------------------
>>Jason Chan (jchan@gmail.com) on 1/10/12 wrote:
>>---------------------------
>>>Bill Henkel on 12/29/11 wrote:
>>>---------------------------
>>>> For desktops and workstations, a differentiator could be
>>>> higher single thread performance rather than more cores
>>>> per socket (e.g. a 6 GHz dual-core water-cooled processor).
>>>
>>>The power consumption of a 6 GHz dual-core x86 processor would not need to be higher
>>>than Sandy Bridge-EP so water cooling would not be necessary. Suppose ...
>>
>>There are probably a lot of issues with this, but one obvious one is heat/power
>>*DENSITY*. Trying to pull 150 watts out of a 30 mm^2 die is probably quite a challenge.
>>
>>I think that the PowerPC 970 had this problem ...
>
>Shouldn't a vapor chamber system with the silicon die exposed (boiling on the silicon surface) work?
Heat pipes have a limit on the amount of power they can absorb in a given surface area of the evaporator. Above that limit, film boiling takes place and cooling fails because only gas is in contact with the evaporator. Going from memory, the best designs are limited to 30 watts per square centimeter while 10 watts is more typical.