Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Matt T

macrumors regular
Original poster
Dec 1, 2005
212
44
Australia
Sorry if this is in the wrong place; I'm a bit new here.:eek:

Does anyone know what the maxium and average TDP of the PowerPC 970 (G5) is?
 
Matt T said:
Sorry if this is in the wrong place; I'm a bit new here.:eek:

Does anyone know what the maxium and average TDP of the PowerPC 970 (G5) is?

"Maximum" and "Average" TDP doesn't make any sense.

TDP stands for "Thermal Design Power". If Intel says a chip has a TDP of 45 Watt (for example), then it means that Apple must design its computers so that they survive and don't die from too much heat, or by destroying the power supply, if the chip uses 45 Watts, 24 hours a day. There is only TDP, there is no average and maximum TDP.
 
Isn't that lower than the Conroe?

How come the G5 PowerMacs have these huge heatsinks over their CPUs then? I always thought it was because the G5s run so damned hot that they need watercooling.
 
generik said:
Isn't that lower than the Conroe?

How come the G5 PowerMacs have these huge heatsinks over their CPUs then? I always thought it was because the G5s run so damned hot that they need watercooling.
Yeah, that perplexes me as well. The 970 has a reasonable TDP when compared to say the Pentium D. (Even lower with the 970FX.) Yet, we see these massive contraptions to cool them in the Power Mac G5.
 
970FXs are more efficient that you would think. A 2GHz 970FX G5 consumes 39 watts maximum. My theory was that Apple didn't do a G5 PowerBook to make IBM look bad before they ditched them.

970FX
 
BlizzardBomb said:
970FXs are more efficient that you would think. A 2GHz 970FX G5 consumes 39 watts maximum. My theory was that Apple didn't do a G5 PowerBook to make IBM look bad before they ditched them.

970FX


PowerBook G5s.....if only they would have been released..... *sigh* :D
 
Two things to realize; 1) IBM's TDP numbers are measured differently from Intel's. Max power is usually about 2x TDP for IBMs. 2) the northbridge on the G5 ran pretty hot due to the crazy fast bus.
 
Catfish_Man said:
Two things to realize; 1) IBM's TDP numbers are measured differently from Intel's. Max power is usually about 2x TDP for IBMs. 2) the northbridge on the G5 ran pretty hot due to the crazy fast bus.

I have never seen the term "TDP" used in any IBM documentation. But I found this IBM webpage

http://www-128.ibm.com/developerworks/library/pa-powerenv/

which shows the power consumption of a 970FX running at 2.5 GHz is 100 Watt. Power consumption can be reduced significantly by switching the chip to half speed or quarter of the speed and reducing the voltage simultaneously, but at full speed it is 100 Watt average.
 
Catfish_Man said:
Two things to realize; 1) IBM's TDP numbers are measured differently from Intel's. Max power is usually about 2x TDP for IBMs. 2) the northbridge on the G5 ran pretty hot due to the crazy fast bus.


Well, that helps explain why the PowerPC G5 is one hot running chip. :)
 
Eidorian said:
Yeah, that perplexes me as well.

It mainly was to keep the system quiet and in the Quads they finally got thing right... they are very quiet systems for the amount of processing power they contain.
 
(repost)

The below lists power consumed by the part, they are not TDP numbers (only part of the power consumed by a chip leaves the chip as heat, heat is what you have to dissipate and is what TDP attempts define).

PPC 970fx power optimized part (@ 2GHz)
40W average, 45-50 W max, 23 W throttle back (half frequency)

PPC 970fx standard part (@ 2GHz)
48W average, 55-60 W max, 29 W throttle back (half frequency)

To me this puts the PPC 970fx below the TDP of a Conroe... I would say the TDP for the PPC 970fx (@2Ghz) is around 40 W (if not lower).

(see section 3.1.5 in this PDF for numbers)
 
gnasher729 said:
"Maximum" and "Average" TDP doesn't make any sense.

TDP stands for "Thermal Design Power". If Intel says a chip has a TDP of 45 Watt (for example), then it means that Apple must design its computers so that they survive and don't die from too much heat, or by destroying the power supply, if the chip uses 45 Watts, 24 hours a day. There is only TDP, there is no average and maximum TDP.

TDP is how much heat the cooling system much be able to remove from the system, it makes no statement about much energy the chip uses.
 
Conroe draws more power than the 970FX, but doesn't generate much more heat, and costs so much less than Yonah I'm sure Apple could afford to give the iMac a bigger power supply and it would still be cheaper than using Merom.
 
GFLPraxis said:
Conroe draws more power than the 970FX, but doesn't generate much more heat, and costs so much less than Yonah I'm sure Apple could afford to give the iMac a bigger power supply and it would still be cheaper than using Merom.
Hence the power supply issue in the iMac and not the thermal dissipation one. :D
 
Most people in this thread seem to be under the assumption that the 65W TDP of Conroe means that the chips have to give off that amount of thermal power. The TDP figure is given for a whole line of CPUs and is used by other manufacturers as a ceiling when they're designing their peripherals. Take the Core 2 Duo E6600 as an example. From the power consumption tests that have been made, it seems to consume less than 50W.

Here's a nice comparison:

http://www.xbitlabs.com/articles/cpu/display/core2duo-shootout_11.html

Also note that those figures don't take efficiency of the CPU power circuits into account, so real CPU power consumption is even lower.
 
Mikael said:
Most people in this thread seem to be under the assumption that the 65W TDP of Conroe means that the chips have to give off that amount of thermal power. The TDP figure is given for a whole line of CPUs and is used by other manufacturers as a ceiling when they're designing their peripherals. Take the Core 2 Duo E6600 as an example. From the power consumption tests that have been made, it seems to consume less than 50W.

Here's a nice comparison:

http://www.xbitlabs.com/articles/cpu/display/core2duo-shootout_11.html

Also note that those figures don't take efficiency of the CPU power circuits into account.
Ah yes, more numbers. Thanks! I like the look of these.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.