Still, it has nothin' on the 603e: typically 4W at 300MHz (back when 300MHz was King o' The Hill), max. 6W .
If I remember correctly the PowerBooks are using the Moto 7447 - According to Moto's specs the 1.3 (used in the 17in) is just over 21 W in power consumption at about 1.6 V.
The new 970FX at 1.4 has that beat. What is really nice is that the package size (25mm x 25mm) is the same. So fitting it in should not be a problem. However there are other design considerations.
Could we see a 750FX PowerBook in a couple of days?
IBM uses its own system controller which might top out at 550MHz bus speed (assuming the usual DDR bus speed doubling).
Any real sources, about this, our just speculation? Though they call it something else, there are no indications anywhere that there are different controllers. IBM would never use apple branded, but printing on chips is cheap and easy I have heard big birdies singing tunes indicating there are no 'IBM' controllers for the 970.
????? what is making everyone think so....I mean I hope so too...but where is the excitement coming from. Macrumors says "something else big" but doesnt elaborate at all. Is this what everyone assumes is a new Powerbook?
I took into account that the 2.0GHz/24.5w ratio increased over the 1.4GHz/12.3w ratio. As I approach 3.0GHz is gets less accurate so...
The 970FX runs with 1V at 2GHZ. It reaches 12.3W at 1.4GHz either because the voltage is reduced to 0.85V or because parts of the execution units are shut off.
And we don't know whether it needs higher voltages to run higher speeds as 2GHz.
So I think your table is interesting but the data basis is weak
For constant voltage and taking 2.0 GHz as a reference:
1.4 GHz --- 17.15 Watts
2.0 GHz --- 24.50 Watts
3.0 GHz --- 36.75 Watts
If 1.0 Volt is normal, reducing it to 0.85 Volts at low clock rate gives:
1.4 GHz --- 12.39 Watts
This works out very close to the published value. The upper table is simply a linear extrapolation from the power published for 2.0 GHz. The lower value is obtained by multiplying the "normal" power at 1.4 GHz by the voltage reduction squared. CPU power typically scales linearly with frequency. Power increases or decreased by the square of the voltage, however.
One other tidbit. If voltage did have to be increased to 1.2 Volts to achieve 3.0 GHz, then power would increase from 36.75 Watts to 52.92 Watt, theoretically.
One thing it means is that I was surprised to see theory work out so close to published values, since other factors affect power. The theory can be thought of in simple terms. For clock rate it is a one to one relationship. Double the frequency and the power doubles. Cut clock rate in half and power gets cut in half.
For voltage it is a squared relationship. Double the voltage and the power goes up four times its value. Cut the voltage in half and the power gets cut to one fourth its value.
Comments
Originally posted by tfworld
Just noticed that the 1.4Ghz 970fx only uses 12.3W!!!
Hmmmm. Chuck one in an iMac, chuck one in a PowerBook, and chuck 2 much faster ones in a Power Mac.
Mmmmm, nice... m.
Originally posted by tfworld
Just noticed that the 1.4Ghz 970fx only uses 12.3W!
It seems the 970fx runs on 0.85V at this speed. So the question is how fast can it go without raising the voltage back to 1V.
Originally posted by Smircle
Anyone notice that the supported bus speed has gone up from 1.0Ghz to 1.1Ghz only?
IBM uses its own system controller which might top out at 550MHz bus speed (assuming the usual DDR bus speed doubling).
Some time ago I read Apples system controller could do 650MHz (maybe be at mosr ) which would fit very nicely with a 2600MHz 970fx...
Originally posted by tfworld
Why did you have to bring up mosr???
Now this entire thread is jinxed. *CONFIRMED* No Powerbook G5 EVAR!1!
Originally posted by Stoo
Still, it has nothin' on the 603e: typically 4W at 300MHz (back when 300MHz was King o' The Hill), max. 6W .
If I remember correctly the PowerBooks are using the Moto 7447 - According to Moto's specs the 1.3 (used in the 17in) is just over 21 W in power consumption at about 1.6 V.
The new 970FX at 1.4 has that beat. What is really nice is that the package size (25mm x 25mm) is the same. So fitting it in should not be a problem. However there are other design considerations.
Could we see a 750FX PowerBook in a couple of days?
if the thing is really going to be as efficient as claimed and only requires 'low power' per se...
that means a larger than 17" screen may become a reality...
perhaps not...
Originally posted by smalM
IBM uses its own system controller which might top out at 550MHz bus speed (assuming the usual DDR bus speed doubling).
Any real sources, about this, our just speculation? Though they call it something else, there are no indications anywhere that there are different controllers. IBM would never use apple branded, but printing on chips is cheap and easy I have heard big birdies singing tunes indicating there are no 'IBM' controllers for the 970.
Originally posted by crobin
IBM would never use apple branded, but printing on chips is cheap and easy
Good objection.
All I've seen is a label. And nowhere a marketing blurb how advanced the IBM contoller is.
So I think you're right and they use the Xserve chip.
Originally posted by smalM
Good objection.
All I've seen is a label. And nowhere a marketing blurb how advanced the IBM contoller is.
So I think you're right and they use the Xserve chip.
When IBM has a shipping computer using their 970 we may know, until then it is all just guess work.
Originally posted by whoami
i love it here when there is good news!
yes, a very nice change from a few years ago.
Originally posted by DMBand0026
**DMBand begins drooling**
mmmmm....g5 power book!
????? what is making everyone think so....I mean I hope so too...but where is the excitement coming from. Macrumors says "something else big" but doesnt elaborate at all. Is this what everyone assumes is a new Powerbook?
GHz---watts
___________________
1.4---12.3
1.6---15.6*
1.8---19.8*
2.0---24.5
2.2---28.8*
2.4---34.1*
2.6---39.8*
2.8---45.9*
3.0---52.5*
* estimated on an increasing GHz/watt ratio
I took into account that the 2.0GHz/24.5w ratio increased over the 1.4GHz/12.3w ratio. As I approach 3.0GHz is gets less accurate so...
Originally posted by Outsider
I took into account that the 2.0GHz/24.5w ratio increased over the 1.4GHz/12.3w ratio. As I approach 3.0GHz is gets less accurate so...
The 970FX runs with 1V at 2GHZ. It reaches 12.3W at 1.4GHz either because the voltage is reduced to 0.85V or because parts of the execution units are shut off.
And we don't know whether it needs higher voltages to run higher speeds as 2GHz.
So I think your table is interesting but the data basis is weak
Although my table is not accurate (it's too linear), I think it's a good gauge, and not too far off.
1.4 GHz --- 17.15 Watts
2.0 GHz --- 24.50 Watts
3.0 GHz --- 36.75 Watts
If 1.0 Volt is normal, reducing it to 0.85 Volts at low clock rate gives:
1.4 GHz --- 12.39 Watts
This works out very close to the published value. The upper table is simply a linear extrapolation from the power published for 2.0 GHz. The lower value is obtained by multiplying the "normal" power at 1.4 GHz by the voltage reduction squared. CPU power typically scales linearly with frequency. Power increases or decreased by the square of the voltage, however.
Originally posted by snoopy
For constant voltage and taking 2.0 GHz as a reference:
1.4 GHz --- 17.15 Watts
2.0 GHz --- 24.50 Watts
3.0 GHz --- 36.75 Watts
If 1.0 Volt is normal, reducing it to 0.85 Volts at low clock rate gives:
1.4 GHz --- 12.39 Watts
What does it all mean Doctor?
Originally posted by Messiahtosh
What does it all mean Doctor?
One thing it means is that I was surprised to see theory work out so close to published values, since other factors affect power. The theory can be thought of in simple terms. For clock rate it is a one to one relationship. Double the frequency and the power doubles. Cut clock rate in half and power gets cut in half.
For voltage it is a squared relationship. Double the voltage and the power goes up four times its value. Cut the voltage in half and the power gets cut to one fourth its value.