Science Project

[carrot]

Hi all
i did i science fair project recently on the amount of heat generated by processors set at different loads and clock speeds. i left tons of holes in it (i didnt say anything about heat management *heat sink*) but that is beside the point.
Anyway, i made it to district and my science teacher was telling me what i need to do to *upgrade* it, if you will. he said that the main thing i need to put on it is how my testing is different from the testing AMD and Intel do on their own in their labs. Well thats all fine and dandy, excpet for one thing. I have no idea what they do in the labs. I was wondering if anyone (saphalline) knows what they actully do.
If not, here is my theory.
in the labs they test the chips up to a temperature that they burn out at. like 90 degrees C. or something. then they do some magic math and come up with a good temperature to run them at. In MY test however. I tested what the temperature is under actual use. so mine is more practical.

I just read over my post. Please Help Me. Please.
i obviously have no idea what i am talking about.

TIA :) :)

[jcnoernberg]

what's your hypothesis ;)?

[Sylvander]

It's important that you learn and understand some fundamentals of thermodynamics.

HEAT & TEMPERATURE are not the same thing; they are totally different.

a. HEAT
...Is the name given to thermal energy while it is in the process of transfer due to a difference of temperature (an indication of the motive force for thermal transfer). [Like water is called rain only when in the process of falling from clouds to the earth beneath due to the force applied by gravity.]
"Heat is an interaction between a system and its surroundings due solely to the temperature difference between them."

b. INTERNAL ENERGY
...Is the name given to this same energy, which is in thermal form, while it is held within some mass of material. [Like clouds which hold water which is not yet rain.]

c. The SPECIFIC HEAT of some material [like a CPU] is a measure of the quantity of thermal energy that will raise a unit of mass of some material by one unit of temperature.

Hence, as thermal energy is generated within the material of a CPU...
given the mass of that CPU, and its SPECIFIC HEAT [at constant pressure] value...
it will take a certain amount of internally generated thermal energy to raise the temperature by one unit [assuming perfect insulation = no heat loss].
Conducting away some of that internal thermal energy in the form of heat will counteract the effect of rising temperature and if the heat loss exactly equals internal energy generated, then the temperature will remain static, neither increasing nor decreasing.
It's a dynamic situation rather like a flow of water into a tank at one end and out at the other. If flow in equals flow out, then the water level remains constant.

d. TEMPERATURE
It's rather tricky to describe exactly what temperature measures.
It is a PROPERTY of a material be it solid, liquid, or gas [its "phase"].
It gives some indication of how much INTERNAL ENERGY is possessed by a unit mass of a specific material in a specific physical "phase".

[jcnoernberg]

it sounds like right now you're doing more of a presentation rather than a science experiment. you need to address some sort of problem, like, "I believe copper is the best heat sink material for cpus". then maybe you can test different heat sink materials on an amd chip and an intel chip, and analyze the results. use a thermal probe running to graphing software on your school computer, and you'll get a neat heat over time graph. you can them educate your viewers on the specific heat of each material and how it corelates to the heat transfer. try copper, aluminum, steel/iron, maybe water somehow, wood, plastic, air! geez the list goes on and on. thermodynamics is pretty interesting, and we're always talking about it on this forum (heatsinks, case fans, cpu temps, etc.)

[Sylvander]

"the amount of heat generated by processors set at different loads and clock speeds"
You'd either need to...
1. Know the SPECIFIC HEAT CAPACITY of a CPU [probably this isn't known] and measure the rate at which internal thermal energy was being generated by insulating the CPU agains heat loss and measure a limited temperature rise [10 or 20 degees Celcius] and find out precisely how long that took...

OR

2. Immerse the CPU in an insulated quantity of a known mass of fluid [water or oil say] with a known SPECIFIC HEAT CAPACITY, then operate the CPU for a precisely measured period of time and precisely measure the resulting final rise in temperature of the fluid. Then using the SPECIFIC HEAT value and mass you calculate the total heat that would produce that effect.
Q = m x CP x [t2 - t1]
where:
Q = heat [Joules]
CP = Specific Heat Capacity [Joules per kg per degees Celcius]
m = mass [kg]
t2 = final temp [degrees Celcius]
t1 = starting temp [degrees Celcius]

Rate of heat output = Q/T [Joules per second]
T = time for that heat to be generated.
1 Watt = 4.18 Joules/sec [if I remember right]

[saphalline]

I agree with jcnoernberg - you have to figure out exactly what you are measuring in relation to a hypothesis. You have to test something, not just show a bunch of data. That's what the district judges are going to look for - can you figure out a meaningful conclusion using the data?

If you so wish to go Sylvander's route of measuring the thermodynamics of CPU's (difficult at best) then you'll need to construct your equipment carefully. And you'll need to do multiple runs of identical set-ups. One of the problems with running benchmarks on a CPU is that the code is not executed the same way every time, so an exact measure of the heat generated by a CPU needs to be averaged out. This is a problem because usually a scientific phenomenon such as this is measured multiple times just because of error in the equipment! If the phenomenon itself is not predictable, it's difficult to measure (thus some of the problems that quantum physicists have to overcome in a statistically relativistic manner).

I can see several directions in which to go for this, but you first need to figure out what you want to measure and what your hypothesis would be.

[carrot]

okay. that was crazy.
I don't think I'm going to do all those thermodynamic things. I don't have enough time/materials. I did however learn something from all that. Thanx Sylvander!
it sounds like right now you're doing more of a presentation rather than a science experiment. you need to address some sort of problem, like, "I believe copper is the best heat sink material for cpus". then maybe you can test different heat sink materials on an amd chip and an intel chip, and analyze the results. use a thermal probe running to graphing software on your school computer, and you'll get a neat heat over time graph. you can them educate your viewers on the specific heat of each material and how it corelates to the heat transfer. try copper, aluminum, steel/iron, maybe water somehow, wood, plastic, air! geez the list goes on and on. thermodynamics is pretty interesting, and we're always talking about it on this forum (heatsinks, case fans, cpu temps, etc.)
Ya, I was kinda out of it when I wrote my first post. I do have a set question though. It is; How do the load and clock speed of a computer affect temperature? My hypothesis is; overclocking the CPU will casue the most amount of heat difference, followed by math operations and finnaly graphic.
my conclusion basicly tol that my hypothesis was wong and that overclocking didnt do as much as setting different loads. (LOL, i didnt even say how much i overclocked!)
Well, I poster way to late and the science fair is already over. but the good news is that I made it to regionals and will be showing it again. so i might still need help.
Thanx for all the help!!!!!

[saphalline]

he said that the main thing i need to put on it is how my testing is different from the testing AMD and Intel do on their own in their labs.Your science teacher doesn't know much about CPU's, does he? :p Normal mortals can't possibly know what AMD and Intel do in their own labs. They keep their secrets well-guarded! Miles of hallways, thick steel doors, multiple security checkpoints, and probably armed guards separate consumers from the inner labs of the CPU giants. The best we can do is guess at what they do in there, although that's not exactly a good method. ;)

They have a good reason for guarding their testing labs so well, however. You know all those core revisions that keep happening over the lifetime of a CPU? Those are the results of all that testing! So you can see why the testing labs are so well-guarded!

Doesn't really matter anyway, since the tests that AMD and Intel do require equipment and knowledge that only the engineers have. No one else has the exact schematic layout of every single transistor. Real hard-core stuff, I tell you...

[carrot]

Thats what i thought. i guessed and put something on my poster. (ill show u if u want, but its at home and im at school)
Well, should i like post a picture of my science fair project and you guys can show me all the huge mistakes i probaly made on it ;)
The regional fair is on march 29th-30th
thanx for all the info!!!!!

[jcnoernberg]

go for it. i used to love the science fair... only in a computer forum could i say that and not be ridiculed! :)

[carrot]

when i get home from school ill see what i can do

[odannyboy000]

cool science project. no one does anything about computers....at least when i was in high school it was like that.

[carrot]

i agree. when i went to look at where i set up my project, it listed everyone and what catagorey they were in. there were like 300 projects and the grand total under the computer science catagorey? 3, including me!
that gave me a slightly higher chance of winning (they pick a certien amount from each catagorey, i think) :D