If geeks love it, we’re on it

Project: Deepfreeze

Project: Deepfreeze


 

 

Ideology – The Foundation:

 

The Cray 2 supercomputer, affectionately named “Bubbles” debuted in 1985 as the ultimate supercomputer of the eighties. NetworkCS didn’t power down the last operating Cray-2 until 1999. Designed from the ground up with the largest amount of memory, the fastest memory, the most unique manufacturing technologies, as well as the fastest processors in the world, the Cray 2 was a technological marvel.

 

The Cray-2, modular up to 8 processors, Cray constructed 31 of these machines:
• Twenty-nine four-way machines
• One eight-way machine
• One single-chip machine

 

The processors employed were Cray’s own 244MHz super-chips. More than twelve times faster than Intel’s fastest experimental chip in 1986, the GaAs (Gallium Arsenide) processor featured the newest and most unique fabrication process in the world.

 

This incredible speed however, came at a significant cost. Producing more than enough heat to give instantaneous third-degree burns, this was the hottest processor on the planet. This heat output is an absolute constant, and was impossible to deal with in conventional means. Therein, the problem was resolved with a new and unique cooling system, officially dubbed a “Super cooler.”

 

The CDC 6600 fluorocarbon condensation unit continues to be the pinnacle of unique, enterprise-level super cooling units. A featuring twin CFC-133 (Freon) cooling chamber, the exorbitantly expensive fluorinert (Produced by 3m) is a bubbling and frothy fluorocarbon (Freon derivative). Electrically neutral by nature, with an ultra-low freeze point, this fluid was prime to be used in a close-loop system.

 

Entering into the insulated copper tubing and copper fluid blocks that rested within the system, the extremely hot processors brought the fluorinert to a boiling temperature, and eventually a gaseous state. Passing through the tubing into the freon chambers, the vapor would condense due to the chilling process, and enter back into the system as an ultra-cold liquid (approximately -121°F).

 

The Plan:

 

I had no such need (Or a wallet) for a device so elegant. In exchange for complexity and cost I employed an ultimately simple, relatively inexpensive, and short-term solution to get the absolute most out of my computer.

 

Riding upon the basic principle of dialectic fluid (That is fluid that is electrically neutral) that exhibits high heat-transfer and dissipation qualities, as well as employing solidified carbon dioxide (Dry ice), I submerged the core components of an overclockable system in an attempt to take it well below 0°F and overclock it.

 

The dry ice, although electrically conductive exhibits the property of “Sublimation.” This is the process of a solid evaporating directly into a gaseous state, skipping the state of liquid matter that could be harmful to the computer itself.

 

Observations:

 

I always found that my processor ran hotter than I would like it. At a mere 1.5v and 2.1GHz, the processor often ran at approximately 95°F. I’m a big fan of really cold processors. The colder the better.

 

Using the same fluid as the Cray-2 used, they submerged a computer into the Fluorinert and then added dry ice.

 

Hypothesis:

 

I hypothesized that the electrically neutral property of OptiCool, and the minor negative charge of frozen carbon dioxide would allow me to successfully submerge and operate my computer.

 

Experiment:

 

To test my hypothesis, I placed my computer into a polystyrene container, pour the liquid on top of it, and then place the dry ice into the container. When the liquid was decently cool, I booted the system and monitor temperatures. When the temperatures were quite low, I planned to overclock the system.

 

 

The OptiCool fluid being poured onto my computer.

 

Conclusion:

 

I brought the dry ice and cold-proof gloves in after my lunch period had ended. Our “Control” group was a small circuit that powered an LED. Pouring a small bit of liquid into the polystyrene box, I threw the LED circuit into the liquid, and it continued to blink showing that it was getting a proper electrical charge. I was extremely nervous.

 

I set the components that I had bravely culled from my crop and slowly laid them in the styrofoam container. The gallons of OptiCool fluid were poured all over the computer components. I had to sigh…More stressed than I have ever been. With heavy-duty gloves, I picked the dual 8 pounds blocks of dry ice out of their containers and placed it into the fluid. Immediately the 75°F fluid started boiling due to the extreme cold of the frozen carbon dioxide (Approximately -158°F).

 

The system was booted with 1.5v. Watching the temperature of the processor at 94°F steadily decrease, I was ready to proceed with lowering the clock speed and booting the system. At 1.33GHz with 1.5v, I knew I could’ve run the processor without a fan.

 

I booted into Windows 2000 and immediately launched a temperature probe. The temperature was steadily dropping as the OptiCool fluid took on the extremely cold temperature of the dry ice.

 

Successfully running a full round of 3dmark2003 (This was before the FutureMark fiasco, and I just ran it to test anyhow), the extreme cold of the system had reduced the electrical resistance in the system, producing the cleanest voltages I have ever had from that power supply. Keeping an eye upon the temperatures, they made their way slowly before the zero mark. Achieving an eventual temperature of negative thirteen degrees Fahrenheit the experiment proved a success.

 

 

An analog alcohol thermometer shows -30°C.

 

 

Sublimed CO2 gas being blown off the surface.

 

 

An ATI Radeon 9700 Pro covered in frost after extraction.

 

Aftermath:

 

In the aftermath, my computer was, and probably still is considerably oily in many places. I’m still surprised that it worked, even despite the theoretical promise of its success. In the process of doing this project, I learned much about dialectic properties of insulating oils, as well as the sublimation process of dry ice. It was an experiment in skill and courage, and I’m glad it resolved positively (Despite the clean-up afterwards.)

 

Materials involved:

 

MSI K7N2 Delta-L:
nVidia nForce2 Chipset:
– A2 Revision
– MCP Southbridge
– ICP Northbridge

 

Processor support:
– Socket A Athlon Classic
– Athlon Thunderbird
– Athlon Palomino
– Athlon Thoroughbred
– Athlon Barton

Memory support:
– PC1600
– PC2100
– PC2700
– PC3200
– Up to 6 banks


AMD Athlon XP 1700+:

“Thoroughbred B” core:
– JIUHB revision
– WPMW Product code
– Produced the tenth week of 2003
– DLT3C code, 1.5v native operation

 

Fabrication:
– Organic packaging
– Nine layers of copper interconnects

 

Speed and specifics:
– 1.47GHz default operation
– 256kb L2 Cache
– 10 stage pipeline

 

Corsair XMS3200c2:
Memory module:
– 6ns winbond chips
– Aluminum RAM sinks
– CAS/RAS/tRP of 2/2/2

 

ATI Radeon 9700 Pro:
Architecture/Bandwidth:
– 8x AGP
– 16 GB/s of memory bandwidth
– 128mb of DDR, 300MHz

 

Supported APIs:
– Up to DirectX 9
– Up to OpenGL 2.3

 

DSI Fluids – Opticool:
Opticool Technical Information

 

Basic dry ice information:

 

Dry Ice is frozen carbon dioxide, a normal part of our earth’s atmosphere. It is the gas that we exhale during breathing and the gas that plants use in photosynthesis. It is also the same gas commonly added to water to make soda water. Dry Ice is particularly useful for freezing, and keeping things frozen because of its very cold temperature: -109.3°F or -78.5°C. Dry Ice is widely used because it is simple to freeze and easy to handle using insulated gloves. Dry Ice changes directly from a solid to a gas -sublimation- in normal atmospheric conditions without going through a wet liquid stage. Therefore it gets the name “dry ice.”As a general rule, Dry Ice will sublimate at a rate of five to ten pounds every 24 hours in a typical ice chest. This sublimation continues from the time of purchase, therefore, pick up Dry Ice as close to the time needed as possible. Bring an ice chest or some other insulated container to hold the Dry Ice and slow the sublimation rate. Dry Ice sublimates faster than regular ice melts but will extend the life of regular ice.

 

It is best not to store Dry Ice in your freezer because your freezer’s thermostat will shut off the freezer due to the extreme cold of the Dry Ice! Of course if the freezer is broken, Dry Ice will save all your frozen goods.
Commercial shipping of perishables often use dry ice even for non frozen goods. Dry ice gives more than twice the cooling energy per pound of weight and three times the cooling energy per volume than regular water ice (H2O). It is often mixed with regular ice to save shipping weight and extend the cooling energy of water ice. Sometimes dry ice is made on the spot from liquid CO2. The resulting dry ice snow is packed in the top of a shipping container offering extended cooling without electrical refrigeration equipment and connections.

 

Final conclusion:

 

An experiment in insanity. Despite my fears, my inhibitions, and completely relearning the old adage about not placing electronics in a liquid, the project went off as a complete success. Courtesy of the Informational Technology Academy, this experiment was even possible in the first regard. Should you have any questions for the author, please email Robert “Thrax” Hallock. Should you have any questions about the ITA program, please email the instructor, Keith Kohring.

 

Comments

  1. Necropolis
    Necropolis Let me be the first to say, What an amazing article and what a brave guy. I would be bricking myself if I had to poor fluid over my motherboard and video card.

    Thrax :thumbsup: Two thumbs up for that one :thumbsup: That is completly off the wall mate.:D
  2. TheLostSwede
    TheLostSwede Friggin awesome Thrax!

    That must have been cool in many ways.
  3. Linc
    Linc Great work, Thrax! :cool:
  4. TheBaron
    TheBaron whoa, that would scare the crap out of me.
  5. Geeky1
    Geeky1 Excellent article!
  6. Thrax
    Thrax Slap this baby up in the article if you'd like, MM.

    Also, video of all of this otw.


    //EDIT:

    One niggling thing that we both missed on the front page:

    "To test my hypothesis, I placed my computer into a polystyrene container, pour the liquid on top of it, and then place the dry ice into the container. When the liquid was decently cool, I booted the system and monitor temperatures. When the temperatures..."

    Should be:

    "To test my hypothesis, I placed my computer into a polystyrene container, poured the liquid on top of it, and then placde the dry ice into the container. When the liquid was decently cool, I booted the system and monitored temperatures. When the temperatures..."

    The tenses were FUBARed.
  7. TBonZ
    TBonZ Nice work Thrax, a very cool read.:thumbsup:
  8. mmonnin
    mmonnin I already PMed Shorty on that Thraxy.;)

    Should have OCed the crap out of it when it was under.
  9. Shorty
    Shorty Il get it updated for ya Thrax!! :D

    Goddamn... this is a fantastic article.. you are nuts! :eek:
  10. Thrax
    Thrax Well, I did OC it, but it only had a k7n2 Delta-L, which has no locked PCI/AGP or anything. So I DID overclock it, that'll be shown when the video goes up.
  11. Shorty
    Shorty Tenses fixed.. where do you want the image placed ?? :)
  12. Thrax
    Thrax Right below the clump of three, if you please. :)
  13. Geeky1
    Geeky1 Little OT but...

    whoah whoah whoah... the K7N2 doesn't have locked PCI/AGP clocks? I thought the nForce2 chipset did that by default... unless MSI is buying special chips from nVidia, how could they unlock the PCI/AGP buses on a chipset that's supposed to lock them? :confused:
  14. edcentric
    edcentric Great Job. Are you working on the desk top version?

    I have run electronics in mineral oil (baby oil), Flurinert, Krytox and PAO synthetic oil. No problemo, as long as you can keep condensation out.

    Did you have any trouble with the fans? They would pull a lot more load in a fluid than in air. That would be my main concern.

    You may need to clean your stuff more often. Now that it is oily it will trap more dust. And there is no solvent for these oils.
  15. Thrax
    Thrax Er, yeah, it has PCI/AGP locks (Sorry, mind-fart, was dealing with someone and their asus kt400 when I was typing the post).

    What I meant to say was that it had an a2 revision and wouldn't go over 180 FSB (I mentioned the A2 in the article).
  16. Thrax
    Thrax All the fans were off the system. The video card's was unplugged, and the HSF didn't even have the F part. ;D

    The OptiCool fluid is great. Wipes right off, and the rest evaporates in a week.
  17. Shorty
    Shorty
    Thrax said
    Right below the clump of three, if you please. :)

    Done buddy :D
  18. Thrax
    Thrax Thanks a ton mate. :D
  19. Geeky1
    Geeky1 I wonder if Mobil1 is electrically conductive... hmm...

    I also wonder if my Grandfather would notice me trying to make off with a 55 gallon drum of dielectric fluid from the EDM department @ the shop... :D
  20. Thrax
    Thrax The Official Project: Deepfreeze Video
    Save target as: 56mb, DiVX 5.0.5, 10:22 in length
  21. maxanon
    maxanon nice project. sounds like a cool prof/program.
  22. Necropolis
    Necropolis
    Thrax said
    The Official Project: Deepfreeze Video
    Save target as: 56mb, DiVX 5.0.5, 10:22 in length

    I try saving the link and its dead. Try left clicking and get a 404 page.
  23. Thrax
    Thrax Works for Shorty, myself, and many others mate. :confused2
  24. Shorty
  25. Necropolis
    Necropolis Ok, New link works:thumbsup:. Old one still doesnt tho. :thumbsdow
  26. Shorty
    Shorty DNS being funky :rolleyes: sorry mate :o
  27. Necropolis
    Necropolis No problem mate. I aint worried, I have it now.:thumbsup:
  28. khan
    khan thrax...you have balls. i woulda used a really old system for that sort of experiment but you used top o the line stuff. bravo to you, that was a great read. now i gotta try it. where did you buy the opticool?
  29. Thrax
  30. Necropolis
    Necropolis I have just watched the video. Jesus Christ that was cold. The layer of frost over everything once you had finished. Your a brave man ;)
  31. Templar
    Templar I wouldn't mind buying a low end PC and trying this over a month or two and see if it still runs. That'd be a lot of dry ice though, and I'd have to find some electrical method of cooling the liquid without letting any condensation near the main components.

    Maybe once I get a decent paying job I could afford this :D
  32. Thrax
    Thrax There is no condensation if it's fully submerged.

    Additionally, the fluid will eventually gel when the temperature gets low enough.
  33. Templar
    Templar I'd think the system would still run even if the liquid is a gel. Perhaps not. :scratch::hrm:
  34. mmonnin
    mmonnin He means the other electrical components that you didnt submerge. You still have to protect those from vapor.
  35. TBonZ
    TBonZ Just watched the video Thrax, mad props to you, you crazy mofo!:thumbsup:
  36. Thrax
    Thrax
    mmonnin said
    He means the other electrical components that you didnt submerge. You still have to protect those from vapor.

    The only vapor being emitted from the dry ice and OptiCool is the sublimation process of the dry ice. The dry ice sublimes directly from its solid state to its gaseous CO2 form. This ambient CO2 bonds instantly with the oxygen particles in the atmosphere and leaves no condensation of any sort.

    Thanks Tbonz! :D
  37. mmonnin
    mmonnin OK that works. What about when you blew on it? No condensation would form?
  38. Thrax
  39. mmonnin
    mmonnin He means the other electrical components that you didnt submerge. You still have to protect those from vapor.
  40. WuGgaRoO
  41. DVDs
    DVDs This has been done 4 years (year 2000) earlier by some crazy people at OCTools.com. They actually did it twice. They used Flourinert, Liquid Nitrogen and Dry Ice. Their project has been copied a lot of times already as well. Good thing though to see a lot of people trying it.

    http://www.octools.com/index.cgi?caller=submersion.html
  42. Thrax
    Thrax Yeah. That's what I based my project off of. I referenced it several times and mentioned it in the documented summary I handed in for the competition which I entered with this project.

    I wanted to see the project myself, and how it would react to a newer generation of HCFCs. It was fun :)

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!