Liquid Nitrogen Meets AMD

Add Your Comments

ED NOTE: Sami sent this in and it’s worth sharing with all – Joe.

Some info of my LN2 experiments with AXIA TB:

I did some LN2 CPU cooling with my week 9 1GHz AXIA TB. Not only did the system work, but I also could break the 2GHz barrier! – stable. :) I think it was also a Finnish national MHz record;)

I could get the chip up to 1470 MHz @1.85V with a ThermoEngine and with my watercooling system (vapor phase cooled liquid -7°C, pics HERE.
I got it up to 1686 MHz, 2.1V rock stable.

At this point, I decided I should test it with LN2. I had never before had any experience with LN2.;) I made an aluminium box and installed it on top of the CPU’s core. Then I booted and went into BIOS to check temps (at this point, there was no LN2 in the box because I wanted to make sure the contact would be OK).

Contact between the CPU’s core and the aluminum box wasn’t that good; temps were over +70°C. Then I did few mods to the mounting system and got a proper contact between them. It was time for LN2;)

I filled the box and went back to BIOS to check temps:
+4°C and dropping. In a few seconds, it was -37°C and I decided to boot into W2K. When I got to W2K and started, the MBM temp sensor displayed 0°C and the reading didn’t change at all. The Temp monitoring chip was probably a bit confused because of the LN2 temp;) [ED: MBM has to be enabled for below zero readings.]

Then I ran some tests and increased the FSB from 133 MHz (1.6 GHz) all the way up to 168 MHz (2016 MHz), the max of my Apacer RAM. The multiplier was set to 12.

At 2016 MHz, I could run tests like pifast or pipih3 without errors (these tests are more “error sensitive” than, eg., 3DMark2000/1, so the CPU was pretty stable). I could run these tests without errors also at 2030 MHz, but the system jammed there; probably my Fortron 350W power supply just couldn’t handle the CPU.

There are cooling pics HERE and
HERE, with benchmarks

Some info about the test system:

  • CPU: 1 GHz AXIA TB week 9
  • Mobo: Abit KT7A (Vcore mod and Vio mod + glygol cooling for chipset)
  • RAM: Apacer PC133 (handpicked 168MHz CAS2)
  • Power: Fortron 350W 32A @ 5V

Just sent this in so that you don’t think Japanese are the only ones with LN2/dry ice cooling methods;)

Sami Mäkinen aka macci

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>