• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

How cold is too cold?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Dawgdoc

Member
Joined
Feb 20, 2007
How does everyone determine where a chip coldbugs?

Is it reasonable to keep chilling the pot until it hangs or restarts when no benching is being performed? Will that give any consistant/reasonable information?

What I have been doing lately is just start benching and experimenting bringing the temps lower and lower until I start to have issues. Not very scientific or organized, but from messing around I seem to be able to figure out what my lowest ideal temps are.

My problem though, is that the temps that I seem to get best results are are significantly warmer than what I see others say works for them.

For example....

My 2 quads (Q6600 G0 and X3220 G0) I didnt really notice any improvements colder than -80C. Both of them appeared to coldbug around -120 to -125C. I see many people benching Q6600s at -140ish.

My E8400 seems to coldbug around -110C and I seem to get best results (only 2 nites of testing so far though) at around -100C.

My "lesser" C2Ds (E4500 and E2160) seems to like the cold more with regards to coldbugging and would be able to run them -135ish, but typically would run them more like -110C or so since I didnt seem to gain anything additional after that.

My s478 CPUs all seem to coldbug around -110C (give or take) and I bench them around -100ishC.

It just seems that most ppl bench LN2 more along the -130 to -140 routinely, and if I just had 1 or 2 chips that did NOT go that low I would say its just those few chips, but I more routinely find my usual benching temp is closer to -100 or -110C and I want to make sure Im not missing anything.

Thanks!
 
While I'm not an expert on extreme cooling I'm pretty sure the there are certain coldbug regions. You said that your chips coldbug around -120º C. Have you tried running them colder? -140-150º C (I know that isn't the more practical thing).

Also, to answer your question about how coldbug temps are found. They are located by trial and error.
 
Hey doc,

The cold bug is like this, when you drop below a certain temperature the system will freeze (in Windows, we know the rest is frozen ;)) and force a reboot - then not POST. The only way to re-start is by letting the container rise in temperature to where it will boot again, you can actually pin it down to a very specific temperature. The tricky part is changing settings can alter the temperature where the bug hits. If you have not seen the board being unable to POST at a certain temperature, you have not cold bugged yet.
 
All so take note that no two chips or boards will be 100% the same when it comes to cold bugs, even if they are the same model.
Some of the top guys will try 20 or 30 chips to find there "killer" cpu, though they only post about that one.

And even when you have a good cpu, the coldbug my change for settings like Maxi pointed out or a bios change or even if you try another board.

And just to make a point of just how nuts it can be, on my IP35pro & QX9650 the coldbug will change by the number of cores I'm using as well as the speed.
It took a hell of alot on LN2 to work it all out!
Now put that same chip into the P5K3 and it coldbugs at -120 no matter the setting, and even then -115 to -120 feels unstable.
But at -110 to -115 I can hit a very unstable 5.8X, something the Abit board cant at that temp.
 
Got it guys.

Looks like I need to just keep pouring that LN2 and experiment and see what happens.

Since the original post, I have been trying colder temps and what I have found is on several of my P4 s478 bench sessions going as cold as -150C gets me SIGNIFICANTLY more mhz on the CPU. On my 3.2EE I was able to add another 600ish mhz going from -100 to -145ish.

Gonna experiment with temps a little more and see how things go. I think many of my previous concepts about what was happening were wrong with regard to temps. I thought I was locking up from the temps, but in many cases I think it was other things.

BTW.....HOLY SMOKES SPi32M and WP1024M take an IMMENSE amount of LN2 to run those benches!!!!
 
You won't be cold bugging a P4 too easily...the newer CPU's (esp 45nm) are much less tolerant of cold)
 
No, cold bugs usually only occur with LN2 and cascades.

oh, ok thanks, I was curious because i have a small fridge in my room and I was thinking one of my crazy plans to somehow convert it into a case :beer:
 
oh, ok thanks, I was curious because i have a small fridge in my room and I was thinking one of my crazy plans to somehow convert it into a case :beer:

Not trying to hijack the thread or rain on your parade, but save your time samurai... fridge coolers are generally doomed to fail since they have no where near the power to cool a rig. I believe theres a topic around somewhere debunking them & explaining why it doesnt work.
 
It will probably end up killing the fridge too, the only CPU that would die in a fridge would be an A64 they can bug really warm....
 
Back