Core i3 thermals...

  • Thread starter Thread starter GT
  • Start date Start date
G

GT

The core i3 chips have graphics chips built into the package now. If a PC
has an external graphics card, I presume it is possible to switch off the
internal GPU, but what does this do to the heat output? For example, the i3
530 is rated at 73w, but with the GPU turned off???
 
The core i3 chips have graphics chips built into the package now. If a PC
has an external graphics card, I presume it is possible to switch off the
internal GPU, but what does this do to the heat output? For example, the i3
530 is rated at 73w, but with the GPU turned off???

Won't Intels web site have some info regarding this ? I've not been
there for awhile but they usually have good data sheets and white papers
etc etc about their products. Might be worth a look.

HTH
bAZZ
 
bAZZ said:
Won't Intels web site have some info regarding this ? I've not been
there for awhile but they usually have good data sheets and white papers
etc etc about their products. Might be worth a look.

Yeah I checked there before posting. All I could find is the total heat
figure, but once you get into the very technical documents, I tend to fall
asleep!!
 
GT said:
Yeah I checked there before posting. All I could find is the total heat
figure, but once you get into the very technical documents, I tend to fall
asleep!!

I had a look at the Intel documents, but I can't figure it out :-)

The i3 has some interesting pins on the interface. It has a
set of GPU_VID signals, which means the graphics core voltage setting,
changes when it isn't rendering anything.

The i3 also puts out a signal, intended to go to the regulator that
powers the GPU, to turn off the regulator entirely. This doesn't sound like
a very wise idea, as you have to worry about feedthru when doing that.
Perhaps the interface to the GPU core is "thin" enough, that it
doesn't matter. They didn't even need to do that, if they didn't want
to, as the all 0's and all 1's values for the GPU VID, can also
turn off the regulator. A separate pin is overkill.

So they could be turning it off, but who can say for sure. A designer
could choose to ignore that signal when designing. The regulator itself
might not even have a disable pin on it. There are plenty of variations.
Any regulator designed to be compliant with the GPU VID table, would disable
the regulator for the all 1's or all 0's values (that is, if the GPU
chooses to output those values, or whatever register is used on the
GPU's behalf).

The GPU clock would appear to run at a constant rate - I couldn't find
a reference to clock switching for it. Many video cards have options for
changing clock rates on demand.

There are a couple current consumption figures for the GPU rail listed
in the spec sheet, but the numbers are quite round and sound like bullshit.
Real numbers would be more random, like 3.14159 amps :-) With the Intel
ampere numbers, you can't really estimate power numbers anyway, because
the associated rail has both VID switching and a load line to consider.
So the number is useless in any case, for what you want it for. It
only helps someone designing a regulator (max amps), and it's hard
to believe in any case.

You can visualize this situation another way though, if you want.
That chunk of stuff, amounts to a Northbridge, and is inside
the CPU. The motherboard only has a Southbridge. In a way, it
is more like a standard architecture, but with the heat from the
Northbridge, being cooled by the CPU heatsink. Turning off the GPU,
would still leave the integrated memory controller and PCI Express
interfaces running. And the memory controller probably has
a pretty hefty power figure.

Paul
 
Back
Top