ICE ICE baby :) (low temperature topic)

[Shodan486]

Do low temperatures (-15 Celsius) increase the computer stability in an OC'd system?

[370forlife]

Re: ICE ICE baby (low temperature topic)

Could. I would be worried about condensation though.

[ratdude747]

Re: ICE ICE baby (low temperature topic)

some have the "cold bug" which could kill a cpu.

http://en.wikipedia.org/wiki/Phenom_II (section "overclocking")

[toastygoodness]

Re: ICE ICE baby (low temperature topic)

you're using liquid nitrogen? that stuff's dangerous, make sure you have real mitts.

i don't think all of the CPU works as a superconductor, so extreme cold probably could kill it. besides, there comes a time when temp simply doesn't matter in an OC. if it was made only to the specifications, you're lucky to even get it to OC. i have two athlon II x2 regor cores, one OCs all the way to 4GHz from 2.6, while the other can only go 2.8. its an arctic cooling freezer64 with mx-2 thermal paste.

[brethin]

Re: ICE ICE baby (low temperature topic)

Long term "Super Cooling" isn't a good idea, its fine for a short period to try and get a really high OC but in general you want a constant temp for long term use. Changing temps cause expansion and contraction and that isn't good.

[Shodan486]

Re: ICE ICE baby (low temperature topic)

I put my PC outside, to my balcony, where it could be like -2 Celsius and the computer acted more stable. So it looks it likes it ...still I don't like the AGP problem - bus frequency raised to 86MHz and its voltage is 3.6v, though cards I'm testing (HD2600XT & N7600GS) only require 0.8v, but anyways the Rialto & HSI chips may go hot very soon, mainly the ATI converter gets pretty hot and eventually hangs up sooner than the GeForce HSI. I only need an 6800 class VGA to prove my self wrong...I'm definitely leaving this problem and solution to mother nature

[PCBONEZ]

Re: ICE ICE baby (low temperature topic)

Yes, colder makes most CPU's faster.
Yes, condensation can kill it.
Yes, heat-up/cool-down cycles can kill it.

We used to dip the end of our cokes in a bucket of liquid nitrogen to have instant coke slushies.
You had to be fast or you'd have an instant coke cube instead.
Yes, use N2 gloves, goggles, and an apron.
.

[Topcat]

Re: ICE ICE baby (low temperature topic)

Seems an awful lot of work just to squeak a couple extra MHz out of a couple cumines on a VP6. I understand the desire to be a little 'different' than all the sheep following the shepherd in the newer hardware realm, and I'm all for experimentation.....but when you're contemplating refrigeration type cooling in attempts to do it, that's going a little too far.....kind of trying to beat a few more miles of travel out of the dead horse....

FWIW, back in the P3 days we tried all kinds of interesting things to OC cumines, including various types of refrigeration. It never worked effectively save for a few benchmarks then kaput.... Water cooling on a lapped cumine was always the best way to keep one cooled and reliable.......and keep in mind, this was before water cooling was just something you could buy. We had to make our water blocks, pumps, and radiators ourselves. My first water cooled system used a heater core from a mazda pickup, a copper block with passages cut into it, and a pair of 120mm 6000RPM A/C powered fans. it was loud and obnoxious, but was effective.

[brethin]

Re: ICE ICE baby (low temperature topic)

Ha those were the days. I skipped water and used a small Dorm (Mini) Fridge as a case, it worked damn nice and condensation wasn't a issue unless I put ice trays in the freezer drawer. :P

[Shodan486]

Re: ICE ICE baby (low temperature topic)

Thanks Top for extending the problematics - the first statement is true, it kinda doesn't mean much to squeeze such a small performance, but this is the how I started, small steps... The shepard theory definitely counts here with me.

And thus the question is: what (a particular component) actually likes these temperatures and helps it to run stable? On the contrary to Topcat's experience, at these temperatures I was able to use my machine for 4 hours without a single reboot, playing games and running everest for 2 hours and that's enough for me. I'm pretty sure it would last for much longer.

[Shodan486]

Re: ICE ICE baby (low temperature topic)

I definitely think it's the HSI / Rialto chip thing. Will have a 6800 series card soon, this should (perhaps) resolve the issue...At least I think it's in this temperature stuff...

[Shodan486]

Re: ICE ICE baby (low temperature topic)

Do you guys think the VP6's chipset heatsink (the black one glued on it) has low heat transfer rate, or any?...Mine is very cool, but I have it under serious airflow. I mean does the paste from the factory working good? I tried to identify the hot spot on this board and the only chip I cannot touch is the chipset, coz it has its heatsink - can I trust it?

[Topcat]

Re: ICE ICE baby (low temperature topic)

^
Nope. those heatsinks with their cheesy sticky tape on the back are hideous and utterly useless. Remove the sink and put something beefier on there. If not going with something bigger, atleast remove the tape and attach it better.

[ratdude747]

Re: ICE ICE baby (low temperature topic)

swap the tape for cpu paste, and glue on a fan.

[Shodan486]

Re: ICE ICE baby (low temperature topic)

Bought this: Click here

I think I can use the ''Few seconds'' glue with a Cooler Master MX-2 cooling compound - that is my last resort and my last suspicion. Will try this ASAP and let you know of the progress, if any.

[Shodan486]

Re: ICE ICE baby (low temperature topic)

Job's done! I think I may leave this board for now, as it seems so far that the cooler helped get rid of the issue - it was the chipset heatspreder's ineffectivness people said, but I thought it was in the glue compound...So I used the one above with MX-2 cooler master (cooling paste) and I could easily play some games at 178FSB but with the easiest memory settings, which suxx...so I stayed with 173,5MHz and the most aggressive memory setting I could research at this speed, which gave me approx 1350MB/s bandwidth over the FSB and DIMMs. Now I'm about to test it on a very unstable and many times patched / drivers install&uninstall XP proffessional and let's see if it ALL was just in the stupid chipset heatsink replacement, which I neglected from the beginning, convincing myself that it could not be a problem at all, coz it was always quite cold, no matter what the FSB frequency...that made me suspicious...now it makes me cry

[Shodan486]

Re: ICE ICE baby (low temperature topic)

So I've tested this system so far and it seems all OK. By knowing that if aggressive timing settings & high clock frequency is bestowed upon the memory modules, I can say that if there is no single error from the start of the computer, no BSOD, no random reboot - there is nothing to worry and only little time needs to be spent for clarifying computers stability status. All of this applies ONLY when every cooling problem is solved - that's what I have just achieved. What I'm trying to say that I just tested as following:

---------------------------------------------
Start
Successful OS start (WIN XP Professional SP2)
Successful Battlefield 2 game start
Successful Loading map
10 Minutes playing the map
End
---------------------------------------------

I know it's ridiculous to test a rig under such pressure by just doing exactly ONLY this, but it has proven to me many times: This game is very ''fat'', it consumes much memory and can engulf all the free space of a 256MB card - that just suxx...

So far I was able to do this 2 times in a row for a 175 & 176MHz FSB/DIMM setting, Memory Timings 2-3-2-6 CAS2, AGP frequency reaches 88MHz and PCI up to 44MHz - I guess this is very nice

I will perform stressful test in the future, not able to test properly so far...

So the final setup is this:

ABIT VP6
Dual 1GHz PIIIE @ 1320MHz Core, 176MHz FSB
2x 512MB Kingston ECC PC133 @ 176MHz CAS2 2-3-2-6
AGP Gainward N7600GS 256DDR3 - GPU:400MHz RAM:1100MHz @ GPU:600MHz RAM: 1500MHz
USB Sound Blaster EAX
Realtek 8129RTLD NIC 100MBit
Chinesse Brand SATA/PATA/USB2.0 PCI
80GB PATA WD
Fortron Epsilon series 700W
Cooler Master HAF 932 Case

The maximum clock setting for the FSB is 178MHz, which requires me to lower from CAS2 to CAS3 and ease the memory timings, which is unacceptable, just FYI.

[NxB]

Re: ICE ICE baby (low temperature topic)

Don't want to be a downer but....

You're putting a lot of work into hardware that can be beaten by a $200 emachine at this point.

The ram in your video card is leaps and bounds ahead of the ram on your motherboard and runs about as fast as your CPU. Not just that, you're missing out an the new instruction sets, 64 bit, etc from the last 10 years.

Even if you got both chips to 2ghz with LN2, they would still be P3s and your ram would still be PC133.

[Shodan486]

Re: ICE ICE baby (low temperature topic)

True point, but that´s not the point. The purpose is to OC a DUAL CPU system, not a single and so far there are few such systems with OC availability, there is a limited list of boards I want to tune up, namely the VP6 (done), Iwill MPX2 (next) and MSI K8N Master2 (when I get one). The last one will be cooled by water, the rest remains air environmental.

Yes, the VGA is a bit leap, but would you believe that it is still slow for this system under these conditions? Radeon 2600XT AGP still performs much better, so it means the AGP bus still has some space left . Next, I may be missing out some instructions and stuff, but I my cpus are for about under 60% constant load, both of them, some peaks, nothing serious.

Well, for games it doesn´t quite matter if it´s dual P3s or Core2 duo, my P3s still got some breath. And my RAM si PC175 right now - symmetric overclocking, pal. And I don´t think that SDRAM would go so much higher.

[AudioCraZ]

Re: ICE ICE baby (low temperature topic)

Condensation is going to be the biggest monster. If you want to keep the cpu happy, and not worry about it, keep it around 20 c. I have ran liquid cooling for years, not to overclock, but to make my system run at optimum performance at standard clocked speeds.