What happens to a cpu with very high Vcore?

[Karsta]

I have read that maximum core voltage of PIII Coppermine is 2.10V. If I use my old cA2 with Vcore 2.10V or over what will happen inside the die? When it fries (HS load temp 34°C) why does it do it? Is more likely to suppose that it just ceases functioning someday or begins to produce errors?

[Jon]

My screen went black and would power up but just a black screen when I did it.

Wasn't because of overvoltage but the CPU was fried nonetheless. Should be the same result. As far as I know there will be no warning signs.

[Karsta]

I have finally downloaded Prime95 and it reported an error quite quickly. I checked the memory (almost 6 h with no problems) with new DocMemory 2.0beta which can be used without dos (means with win ME and 2000). I also lowered temperatures several degrees. Still errors after 1-2 hours. I also installed heatsinks to MOSFETs, PLL and voltage regulator. Raising VIO didn't help (actually vice versa).

What I have left to test is to higher the Vcore even more. Actually I am right now torturing the machine with Prime95 and Vcore 2.10. Core voltage is actually 2.02 to 2.05 under load. Over 2 hours without problems!

[WillysNut]

One of the main problems with high vcore and the associated heat that occurs if not properly cooled is called electromigration. Basically a buildup of heat can cause the interconnects in the die of your cpu to break or become deformed. Once enough of 'em go..so does your CPU. There is a scientific, but understandable, description of electromigration here (http://www.csl.mete.metu.edu.tr/Electromigration/emig.htm). Your load temps look very good so I wouldn't be concerned.

[Karsta]

Thank you for the link!

It is just about what I'm curious about. I read it quite fast through (will use some more time on it later) but it seems that it is local increase in temperature that affects electromigration. So it can be possible that although overall chip temp is low, somewhere inside it can be sufficiently high to cause damage?

[WillysNut]

Originally posted by Karsta

So it can be possible that although overall chip temp is low, somewhere inside it can be sufficiently high to cause damage?

Without a doubt. The floating point unit in most CPU's generates a bit of heat, along with cache. You might have already heard that the early P4's suffered from a mysterious hotspot that led to Intel enabling "clock gating" when the chip got too hot (too hot meaning hotter than most AMD CPU's!!!).

A cDO stepping chip was manufactured after the process for the first generation P3's was mastered. You have a sweet, cool running, very overbusable chip. Enjoy it!!

[Karsta]

Originally posted by WillysNut


You have a sweet, cool running, very overbusable chip. Enjoy it!!

Yes! Right now I'm enjoying a lot. 5 hours of Prime95 without problems! It seems that this chip just needs much. This computer has gotten well enough attention if not else.

I can only hope that it tolerates this 2.10V for longer time (= I'll get a new computer within 6 months and this goes back to 700-770MHz with GORB so that no additional case fans are needed).

I would like to be able to run it hotter so I could cut down the noise from fans.

[Karsta]

That was a bit early:( Prime95 reported error after 5h 35min!

So what next?
I think that computer is stable for all the normal uses, but knowing that it is not fully is frustrating. Should I take the Vcore even higher and try to lower temperature? Feels a bit risky to me.

[outhouse]

Hey Karsta :)

give it a week or two at your new settings so that it has a chance to burn in, when i went to 2.10 2 or 3 months ago i had to run Seti for 2 weeks 24\7 to gain stability. As far as lowering temps it helps me if i get my CPU below 29C at full load i can almost make 1300 but 1292 is stable at 29C

[h2k]

For a couple months before my cel 566 died, it gave divide errors on startup randomly. And sometimes it would just reboot

[Karsta]

Thanks Outhouse :) I needed that encouragement. I can feel happy about my results again and have a clear direction to go.

Now I know that it is my cpu that's problem and that I can still do something about it. I think I'll stick to that 2.10V as you advised and try to improve the cooling even more. I have little problems with the cooling/noise though. Noise is allready too loud for me to keep computer on all the time (I sleep in the same room).

You have gotten that 1gig very far. I hope you will reach that 1300.