Hybrid cooling for gpu

[qiplayer]

Hello, looking to my graphic card I was thinking it should be possible to connect (with an adaptation) the watercooling tubes to the tubes of the cooler of the card.

To make water pass inside the aircooling tubes and give it an additional cool.

Thinking about it later, I thought, the best system to do this would be, making a hole through the plate that's directly on the gpu. And attach to it the watertubes, in the way that the water goes directly where is the hottest place.
Here a picture just to show you where the watertubes should connect to the plate


Uploaded with ImageShack.us


A system like this cud be created by the user on some actual air cooled cards, depends on the thickness of the plate.
And shouldn't be too expensive to add in factory.

So the card that has this system in factory should just have on the side two points where to attach the watertubes. Inside the tubes go through the plate.

In this way if an user has not watercooling in the system, fine, he leave it as it is, and it will cool with the fan.
If he has a watersystem, for example to cool the cpu, he can attach the tubes to the card, to keep it 20 degree cooler.



Have you ever thought about something like this?
If somebody has time and will and the means to do it it would be nice to make an experiment. The problem I see, is in making the hole through the plate, as just looking at it we don't know how it is inside, so something cud go wrong and the plate be damaged.


What do you think about it? would it be a good hybrid system?

 

[Aldakoopa]

I don't think heatpipes and water cooling would flow together too well.

... No pun intended.

 

[ssjwizard]

Not likely. Effective water cooling needs quality blocks designed for exactly that. Effective air cooling is even more so. You would think such a design to be the best of both worlds but infact it would end up being underwhelming at both. This is due to the fact you would have to remove heatpipes to make water channels thereby reducing the air cooling capacity, and the minimal amount of WC contact would make it just as ineffective.

 

[qiplayer]

Not likely. Effective water cooling needs quality blocks designed for exactly that. Effective air cooling is even more so. You would think such a design to be the best of both worlds but infact it would end up being underwhelming at both. This is due to the fact you would have to remove heatpipes to make water channels thereby reducing the air cooling capacity, and the minimal amount of WC contact would make it just as ineffective.

Tank you for the reply

but
man they made it, to reduce the size of chips to 28 nanometers, and you tell me that a hybrid cooling is inpossible, I don't believe it

To make it by factory there cud be different ways. One would be, take the same tubes as you see above, but make them continuous from one to the other like the shape of an S. At one extreme the intake for the water, and on the other side the out. At the extremes there will be a point where the tube makes a T, on one side the heatsink conducts heat to the radiator, on the other side the tube goes out to the connection to the water intake.

The water will not go to the radiator, but as it usually only transport the heat I really can't see this limit you say.

of course to not loose cooling potential by the radiator it must be well projected, and to have enough water cooling potential also.

I would say that the water passing only in one direction through the plate already risks to take away 40% of the heat. The water tube that goes through the plate must be phisically well solded to the rest of the tubes, if it's not, then yes, happens what you say:
In the moment you don't use the hybrid capabilityes you won't have enough cooling in that zone of the plate where the watertube goes through.

 

[wagex]

I could see it working with a little improvement