Cooling Devices Sci. Project

[Firestrider]

I have some questions on computer cooling devices because I'm starting a science project dealing with effects of computer cooling devices on computer performance, advice on how I should start I should start this or go about doing this would be greatly appriciated (sp?). I have some possible questions to this...
How does the temperature of the CPU affect the speed?
What are the average temperatures of fan cooling on PCs?
What other factors effect computer speed?
What are the average temperatures of water cooling on PCs?
Does the room temperature effect anything?
I need to make graphs of the temperature ranges from different devices (fans, heat pipes, water cooling, whatever), the cost of that device, the core speed of your computer, and whatever information that would be needed in this project. This is an average 9th grade project so I'm trying to keep it simple
. Thank you

 

[I.M.O.G.]

First of all, the graphs concerning core speed have a lot more to do with the type of processor being used than any environmental factors... Some chips overclock like the devil and others won't want to budge. I would avoid that specific of a topic.

If you were expecting a rule such as X amount of temperature improvement means a Y amount of CPU performance improvement this will not be possible, unless you choose ONE CPU to base your project on.

A lot of this is my interpretation and may not be the exact truth so further research should be done before quoteing or paraphrasing any of this in your work. This should give you a jump start though.

1) Lower temperature lowers electrical resistance which allows circuits to operate faster. Higher temperatures can cause instability in a CPU, lower temperatures can increase stability and allow for higher performance aka more overclocking.

2) Average temperatures of fan cooling would be in rigs like Dell, Gateway, and such with stock cooling - typical temperatures would be roughly 20-30C above ambient. In enthusiast cooling using aftermarket heatsinks and fans, typical temperatures are closer to 15-25C above ambient with overclocking. (ambient = room air temp) (typical temperatures would be a better word choice than average - average implies that there is some kind of calculation there when that is not possible in reality)

3) Other than temperature, Voltage and the quality of the electrical current they are fed with affect computer speed. The power supply's ability to supply clean stable voltages and sufficient amperage to the components is important for stability. If you want to get into microprocessors here, we could go into manufacturing and how that effects speed but that will get slightly more complicated than you probably want.

4) Average temperature of water cooling is probably around 10-15C above ambient also, not too different than air cooling - but it can achieve the same results much more quietly due to the larger size of the radiator in comparison to the heatsink - so less airflow (noise) is required.

5) Room temperature is the ultimate control. Unless using peltier, bong, geo-thermal, phase change, or sterling cycle cooling you cannot get temperatures below room temperature. The temperature of your CPU is directly affected by the temperature of the room.

 

[creepy]

Room temp effects quite a bit.....in my room i get temps of of about 52ºC cpu and 34ºC nb. When i move my comp. to the game room i get cpu temps of 48ºC and nb temps of 29ºC.

 

[Firestrider]

So I would need to know the ambient room temp., cooling temp., voltage, type of cooling used, and price of the product for my graphs? What would be the procedures used here? What would be the application of this project? Thank you IMOG for all the information.

 

[mccoyn]

Do you mean to measure CPU preformance as it relates to temperatures? This is a common misconception many people have. For the most part, cooler computers don't run any faster unless you activly overclock them. The speed of the computer is ultimatly controlled by a clock generator. That clock generator is ultimatly controlled by a quartz crystal with a very specific size. That quartz crystal doesn't change its frequency (speed) much.

There are exceptions of course. Overclocking is one. If you try to push the processor to the max by changing the clock frequency you can determine the max speed of the processor based on temperature.

Another possible relationship is protective measures built into computers. Motherboards and CPUs will typically slow down if things get really hot. This is just to prevent permenant damage and usally doesn't happen until 60-70 C.

I think to get any useful information you will need to overclock the processor. This can be a lengthy process to find the best possible speed for different temperatures.

 

[Firestrider]

I just want a simple project. Could everyone list what room temperature, the cooling temperature, the voltage, and the type of cooling device they have please?

 

[mccoyn]

If you are looking for numbers, this might be a good resource.

 

[cstarritt]

basicly, the temp of a cpu under load is based on three things
1. the power output on teh cpu (in watts, the amount of heat it gives off)
2. the ambient temperature
3. the degrees C per watt that the cpu rises from ambiant (the c/w rating)

for instance, with cpu giving off 100watts and a cooling device that will give you real world performance of .2 c/w ( a .2 C rise in temp per watt of heat load) you would get 20C above ambient temperature. there is a chart around here somewhere (sorry, dont have the link) that lists the effective c/w numbers for some popular heatsinks.


another thing i should mention is that cooling doesnt itself make a cpu run faster. what cooling does is alow it to run more stable at higher speeds IF you overclock it. more cooling means you can push your cpu farther.

if looking for a good example of a overclocking chip, take a look at ppl using a 2.4c p4 or an athlon Xp1700+ JIUHB.
from the main overclockers.com site you can veiw the cpu database and take a look at some results.

 

[vincent_1985]

Once your done I wouldn';t mind you posting your project so we can see all yoru nice work! =)

 

[Firestrider]

Here are some of my friends and my computer voltage and room temps... is there anyway to calculate the processor temperature from this?

Voltage 1.568v 3.3v 2.0v
Room Temp. 73ºF 65ºF 72ºF
Processor Temp 26ºC n/a n/a
Cooling Fan Fan Fan

I ran a program called PC wizard 2004 to find my processor to find my processor temp. but it had no reading with the same program on my friends computers. I was planning on making a graph on just this.

 

[Smirabi]

What would be the procedures used here? What would be the application of this project?

thats a tough one. at this point in time, none. other than for your own knowledge (understanding what factors affect your cpu temp), there is no functional or shareable gain here.

this brings up a number of problems with how temps are reported, especially in reviews of blocks and heatsinks. yeah, sure, everyone who writes reviews does "multiple" mounts and calculates an average. but if you want to be able to say X block is better than Y block (or heatsink/fan), you need to use statistical tests of significance.

this might belong somewhere else, but it does apply to what you will do: never report that one block is better than another just by looking at average readings over a few mounts. you can say that results suggest Y is better than X, but to conclude that Y is better than X, you need to use statistics (like a t- test, or at least a Mann-Whitney U).

if you want to be able to report with surity, PM me and i can tell you how to do so.

good luck.

 

[I.M.O.G.]

I just want a simple project. Could everyone list what room temperature, the cooling temperature, the voltage, and the type of cooling device they have please?

Let me be more clear, that is a BAD IDEA for a project. I don't recommend it at all.

What I was trying to say earlier is that you cannot compare these readings with different CPU's and Motherboards. When put side by side, the numbers are meaningless - there is nothing you can compare or say anything about.

Maybe the people at the science fair won't be smart enough to know this, but maybe someone will.

I recommend you don't involve temperature at all, because doing so will make your science project bogus basically. I think you should just run a benchmark program that displays your memory bandwidth and graph that at different settings. These values are correct and accurate. By changing the FSB you alter your memory bandwidth, so you can compare how changes to the FSB effect memory bandwidth if you want. This can be done with a single computer using these different FSB settings, or it can just be done with different computers and you can note their hardware - so you could compare your computer's memory bandwidth with that of your friend's.

Sisoft Sandra is a program that will display graphs of your memory bandwidth. It is free I believe. Click here to be taken to a page where you can download it.

If you want to keep things simple and correct, you do not want to use temperatures. Memory bandwidth would be much easier and more importantly for a science project, it will be correct.

The procedure of this project would explain how you went about attaining the data.

The application of this project would explain how the results of using the procedure explain performance differences between different settings and/or hardware.

If you choose to do your project this way, I will be sure to check on this thread daily atleast and help you with any questions, and I'm sure everyone else will help you also. I can't help you much with the temperature thing though because there is too much wrong with trying to compare or say anything about temperature readings from different computers.

Here are some of my friends and my computer voltage and room temps... is there anyway to calculate the processor temperature from this?

Voltage 1.568v 3.3v 2.0v
Room Temp. 73ºF 65ºF 72ºF
Processor Temp 26ºC n/a n/a
Cooling Fan Fan Fan

I ran a program called PC wizard 2004 to find my processor to find my processor temp. but it had no reading with the same program on my friends computers. I was planning on making a graph on just this.

No you cannot calculate temperature accurately - you could get a very poor calculation using fairly complicated mathematics though.

In order to get temperature readings, the computer has to have temperature sensors - many computers do not have temperature sensors. That is likely why your friends computer gave no reading.