Hello
it may be silly to ask this question ? but it may be a good tip
i've got an asus board with a intel cpu 3ghz model 630 , accroding the intel web site this cpu takes 85 watt
I've got the idea to reduce cpu frequency to 2 ghz using AiBooster utility, will it reduce power consumption
hicham
reducing cpu frequency does it reduce power consumption ?
Moderators: Ralf Hutter, Lawrence Lee
-
frostedflakes
- Posts: 1608
- Joined: Tue Jan 04, 2005 4:02 pm
- Location: United States
Yes there is a linear relationship between power consumption and clock frequency, meaning a 33% reduction in clock speed should bring CPU power consumption down to about 57w. 
And if you have the ability to do so reduce voltage as well, as it has a quadratic relationship with power consumption, i.e. a 25% reduction in voltage will reduce power consumption by about 50%.
And if you have the ability to do so reduce voltage as well, as it has a quadratic relationship with power consumption, i.e. a 25% reduction in voltage will reduce power consumption by about 50%.
Hi hichamlin,
The answer is yes, but not quite as simple as frostedflakes stated.
First of all, the 85 watt rating is a guideline for the thermal design for your system. A specific part may hit 85 watts, but that is very difficult if not impossible to sustain. (But if your cooling can handle 85 watts, you are safe.)
You will notice that some whole families of chips use that same rating, so if you have one of the slower chips in the family, and happen to get one that runs cool, your particular processor could be well under 85 watts.
A portion of the power usage in the processor is consumed by the transistors switching, and that portion of the power usage is proportional to the speed, but there are other consumers of power in the chip too that are more fixed at a given temperature and voltage (resistance and leakage current).
In general, lowering the voltage, lowering the temperature and lowering the clockrate all will reduce power usage. ( And there is even a little positive reinforcement here, as lowering the clockrate may allow the power to be reduced a little more.)
The answer is yes, but not quite as simple as frostedflakes stated.
First of all, the 85 watt rating is a guideline for the thermal design for your system. A specific part may hit 85 watts, but that is very difficult if not impossible to sustain. (But if your cooling can handle 85 watts, you are safe.)
You will notice that some whole families of chips use that same rating, so if you have one of the slower chips in the family, and happen to get one that runs cool, your particular processor could be well under 85 watts.
A portion of the power usage in the processor is consumed by the transistors switching, and that portion of the power usage is proportional to the speed, but there are other consumers of power in the chip too that are more fixed at a given temperature and voltage (resistance and leakage current).
In general, lowering the voltage, lowering the temperature and lowering the clockrate all will reduce power usage. ( And there is even a little positive reinforcement here, as lowering the clockrate may allow the power to be reduced a little more.)
-
frostedflakes
- Posts: 1608
- Joined: Tue Jan 04, 2005 4:02 pm
- Location: United States
True, as with about anything you can make it as complicated as you want. 
There are different types of loads in a CPU (static and dynamic) and clock speed only effects power consumption due to dynamic (i.e. switching) loads. Nonetheless, the linear relationship is a good approximation, or at least good enough for our use.
But you are right about the TDP, it should be mentioned that it is not necessarily indicative of real-world power consumption. SPCR measured the 630 Prescott to use 78w including VRM losses, so load CPU power consumption is probably in the 60-65w range.
http://www.silentpcreview.com/article313-page5.html
There are different types of loads in a CPU (static and dynamic) and clock speed only effects power consumption due to dynamic (i.e. switching) loads. Nonetheless, the linear relationship is a good approximation, or at least good enough for our use.
But you are right about the TDP, it should be mentioned that it is not necessarily indicative of real-world power consumption. SPCR measured the 630 Prescott to use 78w including VRM losses, so load CPU power consumption is probably in the 60-65w range.
http://www.silentpcreview.com/article313-page5.html
Does the answer to this depend on how busy the CPU is?
In the extreme case, if it is busy all the time, then reducing the clock rate means it has to run longer to finish the job, so it may take more total power.
(even if the power use at any one time is reduced).
If it is idling most of the time, does the idle power depend much on the clock rate? (Does it take much more power to run HLT instructions at 3GHz than it takes at 2 GHz).
--
We're going noplace mighty fast - Scotty.
In the extreme case, if it is busy all the time, then reducing the clock rate means it has to run longer to finish the job, so it may take more total power.
(even if the power use at any one time is reduced).
If it is idling most of the time, does the idle power depend much on the clock rate? (Does it take much more power to run HLT instructions at 3GHz than it takes at 2 GHz).
--
We're going noplace mighty fast - Scotty.
Hello guys
thanks for replying ,
I'm stil a noob in the overcloking cpu field
what i did is just reduce the external frequecy parameter to 150 in AiBooster
cpu voltage is set to auto, this brought the cpu frequency to 2310 mhz
a link to a tutorial or guide ?
and most of time, I'm using the pc to browse the net and emails, so most of the time the cpu is so busy,
thanks for replying ,
I'm stil a noob in the overcloking cpu field
what i did is just reduce the external frequecy parameter to 150 in AiBooster
cpu voltage is set to auto, this brought the cpu frequency to 2310 mhz
a link to a tutorial or guide ?
and most of time, I'm using the pc to browse the net and emails, so most of the time the cpu is so busy,