<font color="cyan">Why?</font> </font>[/QUOTE]why not? its his machine, time and effort [img]smile.gif[/img] give him some slack.

Physics, shmysics. It's submerged. It looks awesome. AND NOW YOU SAY IT COULD NOT POSSIBLY WORK!? The universe better stop staring at its own belly button and rearrange itself to accommodate this. And raise the speed of light while it's at it. And make space habitable.

I guess you've never heard of

a) cars without radiators (VW bug and Porche 911 prime examples)

b) convection cooling

This works. The oil cools itself with convection airflow across the top of the oil. The oil cools the parts by convection flow across the hot parts. (you can see this in the video - look at the "heat waves" moving in the oil)

They ran 3Dmark05 - that stresses the machine as much as Oblivion or any other game.

BTW, thermal equilibrium was reached at 40C - that is to say, when the temperature reached 40C it didn't rise any further. This is better than my machine running with enhanced cooling - it runs at 45C.

Fans won't work, but not for the reasons you specify. Fans won't turn very fast in fluid when they're designed to turn in air - too much drag. This will ramp the current requirements up significantly. In fact it will behave much like a short circuit and the fans will likely burn out.

In fact, if you'd read the article you'd know all these things.

Tell me, Mr.Engineer man:

1) how many watts does the computer in the article dissipate?

2) What is the limit (how many watts) can oil cooling dissipate to air without a radiator?

Demonstrate how the answers invalidate the experiment shown on Tom's Hardware.

Provide proof, not conjecture.

<font color="cyan">Firstly the article didn't load for me, so I never read it. I watched a video thats it.

I still stand by my statement, it won't last.

On the topic of cars, Old Porsche's and VW Beetles are air cooled. That computer still isn't air cooled.

I don't believe that convection through a plastic case is enough. Or even a glass case. After time, I still believe that it will heat up to the point where the cooling is lost.

And i still stand by the fact the only way to get that running properly is with a cooling system.

Plus if it's only air cooling with motionless oil, the key being motionless, I don't think it will dissapate enough heat.

And don't slate my credentials, I work on 66,000BHP engines. I know about physics and heat. I had to study thermodynamics at college.</font>

DINGDING! Correct, the vw/porche is aircooled - no radiator - but, SURPRISE the computer is aircooled. All cooling systems ultimatey rely on transfer to air as their cooling medium, whatever intermediate means they use.

I've already answered your question. Now answer mine, or admit defeat.

:rolleyes:

It does have a cooling system - a convection cooling oil bath. If you'd read the article, you'd know that the primary cooling transfer isn't through the case - it's to air at the top.

The oil isn't motionless, hot oil rises, cool oil drops - the exact same behaviour you see with air, water, or any other medium used to cool something.

Now, I'll get more specific with the questions in an attempt to lead you back to reason:

1) Specifically how many watts/cm^2 can air cooling remove from oil (or any other surface, such as an engine block)?

2) What is the total heat output of the computer used to test the oil bath?

If you can't answer both these questions, then you cannot refute me here.

[ 09-14-2006, 07:46 PM: Message edited by: Sir Krustin ]

<font color="cyan">Why?</font> </font>[/QUOTE]Because it will make me happy.
Why does everyone think that everything should have a reason?

<font color=8fbc8f>Vaugly reminds me of a fartcan on a pink car I've seen somewhere recently. Some people!

Just one minor detail to mention, a normal CPU will run happy at 104*F /40*C. If doing this at home be sure to read the last line and use the recommended liquid of choice, "motor oil". It will hardly have any heat strains on it what-so-ever, and will last for hours without needing a scheduled change out, or a reminder note from the local pit-stop. [img]smile.gif[/img] </font>

<font color="cyan">Why?</font> </font>[/QUOTE]Because it will make me happy.
Why does everyone think that everything should have a reason? </font>[/QUOTE]That'll teach you to ever make anything again and speak about it! :D

<font color="cyan">Ok, so the oil is moving in a cycle and cooling to the air at the top. Still doesn't sound to me like it will do enough cooling.

And a Porsche/Beetle has air flow across the engine, which is how it's cooled. Just seems to me that the air in this computer case will be stagnant, and not get rid of enough heat.

And i don't know the exact figures your asking me, so I can still have an opinion [img]tongue.gif[/img]

Even if I am wrong lmao!</font>

Certainly you'd want to use a light oil... but with a temp differential between air and cpu of 30+ degrees you'd definitely get plenty of circulation.

If you knew the ambient temp when toms ran their test you could do a seat of the pants calculation to figure out how close to the limit they were at full load.

The only thing I'd change in their design is that I'd seal the top and use a heat exchanger to extract the heat. A couple aluminum radiators with copper heat pipes would do the trick nicely. I think you could find a table that would tell you how big the air side radiator would need to be to dissipate a couple hundred watts of heat. The oil side radiator would be relatively smaller obviously and I imagine there's a table somewhere for those too.

[ 09-15-2006, 11:18 AM: Message edited by: Thoran ]