Aside from a little extra signaling and bus overhead, the number of attached devices is mostly irrelevant.
So you're claiming that the overhead of the SATA logic is an order of magnitude more than 0.2 Watts ? How much ? 2 Watts ? 20 Watts ?
Let's do a little math with those TDP figures ...
The
total draw for the
old ICH8 chip with the following logic: DMI, PCIe, PCI, SATA, USB and HD Audio =
3 Watts
But this is with 4 SATA devices attached. So let's remove that and we have a 3.0 - (0.2 x 4 ) = 2.2 Watts
Notice there are 5 devices also attached to the PCI (3) and PCIe (2) buses. And 8 and 2 devices attached to the USB bus. But I'll tell you what ... let's just subtract 0.1 Watts per PCI/PCIe device for argument's sake and ignore the USB ... so that's 2.2 Watts - 0.5 Watts = 1.7 Watts ( You can see that the power draw of the chip jumps from 3.7W to 4.1W when you attach a x4 load, let's assume every x1 load causes a 0.1W jump in power )
So the OVERHEAD of the logic itself is around 1.7 Watts ... and that logic includes the DMI 4x switch, the PCIe logic, PCI, SATA, USB and HD Audio
Let's further say that USB and HDAudio don't contribute anything here. We're still left with DMI, PCIe, PCI and SATA. Now the DMI 4x logic is the most complex and runs at the highest speed of the four. And PCIe is more complex and runs at higher speed than PCI, which itself runs faster than SATA. So it would make sense that DMI logic would take significant percentage of the power.
But never mind, let's assume they all take equal amounts even though the SATA logic should take the smallest percentage of this. So that's 1.7 Watts / 4 =
0.425 Watts. At load.
And this is for a chip that's a generation behind in process technology than the GF9400m. You surely don't think the 9400m's 65nm tech SATA logic has the same power draw as the ICH8's 90nm tech SATA logic do you ?
Yes I haven't considered SATA 1.5 Gbit/s, so let's consider. How much savings do we get when we downclock the SATA logic ? Half ? A third of the normal power required for 3.0 Gbit/s operation ? Even saving at a third ..that's 0.14 watts. So we save ... 0.285 Watts.
So for paltry 0.3 Watts we would rather downclock the SATA ?
Unless of course you're telling me that the SATA logic itself is so complex that it takes 2 Watts out of the 2.2 Watts. And that PCI, PCI express, and the DMI logic is of no consequence even though it is far more complex and operates at much higher speeds. Do you realise how ludicrous you would sound ?
Overall power savings for this single design change may be small, on the order of 1W or less, but if it's one piece of an overall efficiency campaign, they start to add up quickly.
So now you're claiming that SAVINGS are > 1W ? Where did you get this figure from ? Is that savings under load ? You have specs for the 9400m ? You did some back of the envelope calculations with comparable technology ? Or processes ? Or like someone else said it's another "argumentum ad ignorantiam " ?
This seems at the very least plausible to me given that the improved battery life these machines are exhibiting seems to be far surpassing what can be explained through the increased capacity of the batteries alone.
And again you conflate your argument with something else. Does the 1W saving, by your admission, represent a
significant percentage of the longer battery life performance ? Yes or no ?
As others have pointed out ... you make absurd statements, improper analogies, contradict yourself, paint yourself into a corner and then ...
Sometimes knownikko it is better to keep quiet and let people merely suspect that you're not the sharpest tool than flaunt your "knowledge" and remove all doubt of it
2x