• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

CPU speed vs. Heat

S

ScanIAm

n00b
Joined
Apr 5, 2004
Messages
6
I have an AMD XP 2600+ and it's been running hot for the last few weeks. At Idle, it's 61C, but right now, it's running at 67C so I'm going to lapp it. Further, I was one of the suckers who picked up bogus thermal paste so I'll be fixing that as well.

My question is, does a higher amount of heat cause the chip to run slower? I know that by cooling the chip, you can safely run it faster, but that is if you make the effort to do so. Does temperature affect the speed of the chip? (and, yes, I realize that if the temperature gets too high, the chip dies...speed=0)
 
Originally posted by ScanIAm
My question is, does a higher amount of heat cause the chip to run slower?
Click to expand...

It's been a while since I took my VLSI fabrication class, but if memory serves: as the temperature increases the atoms of the semiconductor (and those of the gates and other components) gain energy. Long story short, if the temperature gets too high the electrons that run through the gates on the semiconductor have an easier time zipping about into places where they should not be. So higher heat may not necessarily mean slower speeds, but it will mean more errors.

If anyone knows more about the actual theory behind semiconductor materials science, please feel free to correct any technical errors I made. No doubt, there are some. As I said, it's been a while since I took a VLSI fab class.

But on the topic of providing a more effective thermal solution: yes, lapping the base of the HSF is a good way to improve the cooling of the CPU. However you must make sure that the lapped surface is as close to perfectly flat as you can manage. Smooth does not necessarily mean flat.

And it is possible to lap a CPU core. It is not recommended, but it is possible.
 
thats what you get for being cheap.

AS3 doesn't cost a lot at all...

i still use AS3.
 
Originally posted by AggieMEEN
It's been a while since I took my VLSI fabrication class, but if memory serves: as the temperature increases the atoms of the semiconductor (and those of the gates and other components) gain energy. Long story short, if the temperature gets too high the electrons that run through the gates on the semiconductor have an easier time zipping about into places where they should not be. So higher heat may not necessarily mean slower speeds, but it will mean more errors.

If anyone knows more about the actual theory behind semiconductor materials science, please feel free to correct any technical errors I made. No doubt, there are some. As I said, it's been a while since I took a VLSI fab class.

But on the topic of providing a more effective thermal solution: yes, lapping the base of the HSF is a good way to improve the cooling of the CPU. However you must make sure that the lapped surface is as close to perfectly flat as you can manage. Smooth does not necessarily mean flat.

And it is possible to lap a CPU core. It is not recommended, but it is possible.
Click to expand...

I'm a materials science & engineering student, but i've only taken one class in electronic materials... I think you're pretty much correct however. Higher temps means the atoms vibrate faster and with greater amplitude (the reason for thermal expansion), therefore they are more likely to get in the way of electrons traveling through the lattice. This basically just changes resistance, and very very slightly affects the average electron speed. But throughout the temperature range that we're talking about here, the difference is very small and I doubt there is any real difference in processor performance.
On a side note, decreasing the temperature of a semiconductor actually increases its overall resistance. The thermal energy is actually needed to give electrons some of the energy required to jump the bandgap and make it through the device. It might also be needed to allow electron/hole pairs to move through the lattice itself... My memory needs a little refreshment, it's been a while since I took that class too.
 
Originally posted by AggieMEEN
It's been a while since I took my VLSI fabrication class, but if memory serves: as the temperature increases the atoms of the semiconductor (and those of the gates and other components) gain energy. Long story short, if the temperature gets too high the electrons that run through the gates on the semiconductor have an easier time zipping about into places where they should not be. So higher heat may not necessarily mean slower speeds, but it will mean more errors.

If anyone knows more about the actual theory behind semiconductor materials science, please feel free to correct any technical errors I made. No doubt, there are some. As I said, it's been a while since I took a VLSI fab class.

But on the topic of providing a more effective thermal solution: yes, lapping the base of the HSF is a good way to improve the cooling of the CPU. However you must make sure that the lapped surface is as close to perfectly flat as you can manage. Smooth does not necessarily mean flat.

And it is possible to lap a CPU core. It is not recommended, but it is possible.
Click to expand...

You're dead on. The theory is called electromigration.

http://www.csl.mete.metu.edu.tr/Electromigration/emig.htm

Matt.
 
Originally posted by enraged78
You're dead on. The theory is called electromigration.

http://www.csl.mete.metu.edu.tr/Electromigration/emig.htm

Matt.
Click to expand...

Yes, but electromigration is usually a gradual process. After time, more and more atoms get knocked out of a certain spot to a spot further "downstream". Finally the original spot is so much smaller in cross-sectional area that current density increases enough to overheat the area and melt it apart... Right?
 
Originally posted by zer0signal667
Yes, but electromigration is usually a gradual process. After time, more and more atoms get knocked out of a certain spot to a spot further "downstream". Finally the original spot is so much smaller in cross-sectional area that current density increases enough to overheat the area and melt it apart... Right?
Click to expand...

It was my understanding the the original spot never increases it's cross sectional area (That's what you mean by smaller, right? Material moving downstream like a river? ). The material removed by the bombarding electrons just allows the electrons to 'jump' through the semiconductor path laid before it to a path of lesser resistance. I'm pretty sure that the removed material does not 'go downstream', it just gets pushed aside, like an arc burn. As more electrons arc to the new path, it becomes larger and larger due to heat and current. It's been about six years since my DSD class, so it's very possible I'm a little off. If anyone has anything to add or correct, please feel free.

Thanks,
Matt.
 
Originally posted by enraged78
It was my understanding the the original spot never increases it's cross sectional area (That's what you mean by smaller, right? Material moving downstream like a river? ). The material removed by the bombarding electrons just allows the electrons to 'jump' through the semiconductor path laid before it to a path of lesser resistance. I'm pretty sure that the removed material does not 'go downstream', it just gets pushed aside, like an arc burn. As more electrons arc to the new path, it becomes larger and larger due to heat and current. It's been about six years since my DSD class, so it's very possible I'm a little off. If anyone has anything to add or correct, please feel free.

Thanks,
Matt.
Click to expand...

It sounds like you're saying that the material being moved is semiconductor material. I was under the impression that electromigration occurred in the metallic traces. By "smaller" I meant the cross sectional area of the trace is decreasing as atoms are being knocked out of the section and moved further down the trace. The smaller cross section area would increase local resistance, and heat, and finally destroy that section of the trace.
I remember seeing a picture of a trace that had been destroyed in such a manner, but maybe I'll go look around and see if I'm wrong...
 
On a side note, the issue was the stock pad/whatever that came with the cooler (Thermaltake Volcano 10+). Put some ceramique on and it's running at 57C. I'll be lappin that puppy this weekend.
 
Oh, I could be wrong, but my impression of the particular color of that OCZ crap is that maybe it uses aluminum in place of the silver. Not that they couldn't get that look through other cheating methods, but the particular color and shine make me think that. Oh well, either way people are getting ripped off and I'm pretty sure it's illegal to do that as well.

BTW, I noticed pads don't take extreme heat well. It took me forever to figure out why my video card was unstable at stock speeds after I oced it until it crashed a few times, then I finally looked at it and realized the core had all but melted through the thermal pad just from a small oc... d-: I bet factory pads are HORRIBLE for ocing with CPUs/etc too.
 
Originally posted by zer0signal667
It sounds like you're saying that the material being moved is semiconductor material. I was under the impression that electromigration occurred in the metallic traces. By "smaller" I meant the cross sectional area of the trace is decreasing as atoms are being knocked out of the section and moved further down the trace. The smaller cross section area would increase local resistance, and heat, and finally destroy that section of the trace.
I remember seeing a picture of a trace that had been destroyed in such a manner, but maybe I'll go look around and see if I'm wrong...
Click to expand...

We're both saying the exact same thing here. The semiconductor path is in between the metallic traces. As the traces get eaten away, the electrons jump off of the semiconductor path, through the old trace, and to the new semiconductor path.

Thanks for clarifying what you meant by smaller cross sectional area. Makes sense now.

Thanks,
Matt.
 
You must log in or register to reply here.
Back
Top