water cooling questions

[Fullmetal Chocobo]

Would water cooling be more effective in a situation with higher ambient temps, versus air? If the ambient temp was say 100+ degrees, and you were trying to cool lower end P4s (say like 1.4GHz P4s), would air be as effective, or less effective than a water cooling setup? This wouldn't be for OCing or anything--just keeping the processors / machines operational.

Also, from what I can see, it seems most WC setups have it setup as follows: blocks -> resevoir -> pump -> radioator -> back to blocks.

If you were using multiple blocks to cool ~4 - 6 processors, could you daisy chain the blocks, or would other modification to the cooling flow be required? Such as adding another pump & resevoir or pump & radiator inbetween?

 

[Nirach]

Daft question but why would you be cooling 4-6 processors? Server or something?

I dunno, depends how hot they got with air cooling as to whether you'd see any benefit from water. Obviously the warm water is going to be going from one block to the next if they're one after another, which would reduce cooling efficiency, especially by the end of the chain.

If I was going to do that, and I had the space, I'd have a radiator between every two CPU blocks, a single 120 or something. And one big-ass pump.

However, in that scenario, I don't believe that water would be more beneficial than air.

My loops both go Res.-Pump-Block-Rad-Res.

The downside to water is there's less airflow in the case, thus, less passive cooling on things like mosfets/ram/passively cooled NB/SB, or anything else passively cooled.

 

[Fullmetal Chocobo]

Well, it is for this project. So space / airflow is not a concern. And there will be 4-6 computers when I am done, hopefully. And as this is in my garage, cooling is going to be interesting aspect of the project. Fortunately, I'm not OCing or going for ultra low temps or anything--I just want to keep the computers operational.

 

[Luckyboy1]

Yes, water cooling would work better.

You would need to have a very large external radiator with fans on it. What I mean by large is at least one of those radiators used as auxillery transmission coolers for medium duty trucks. The big and blocky kind and NOT the cheaper ones for auto use. The big and blocky types are big enough to take plastic wire ties and mount four, 1200 mm fans on them.

With the P4's at less than 2.0 GHz at stock speeds and voltages, you'll get about 2 C gain in temp between blocks. So, if the room temp is 100 F, which converts to... lemme see how many Commies that converts into! that's slightly less than 38 C. Let's say you've got great radiator capacity... not some little radiator that will fit inside the PC, I mean decent radiator capacity!... Your water under those conditions would probably come out of the radiator at 42 C. Add 12 C for 6 water blocks and you are still at 54 C, which would be decent enough for stock applications.

The problem would be not just in having a big enough radiator. You would also need a pump with enough head pressure as well as flow rate to push through that long loop. Still, pumps likle that are available and if nothing else, you can get a pump designed to run baseboard water heating systems.

 

[Nirach]

Well. If you did go automotive radiator route, there'd probably be some modification for the barbs, although, I'm sure that'd not be beyond you. I'd be somewhat uncomfortable having just one rad in a loop that size, even with an automotive standard radiator. I'd still want something between every two/three, so.. Twoblocks-rad-twoblocks-rad-twoblocks-largerad/threeblocks-rad-threeblocks-larg
rad to at least make an effort to keep summer temps managable too. I'd say you might want something bigger than what people put in their PC, pump wise. I'd take a look around at the large out door pond pumps, or what Luckyboy mentioned. Weight on the tubes might also need to be considered, possible kinks etc.. For the larger, not so mobile parts, maybe some coper pipe that you can fit into 1/2" tubing, as it's copper and not PVC, there'd not be so much lost in wall thickness. Wormclip, or ziptie, the connections to keep them water tight.

 

[Luckyboy1]

Oh yeah, and I forgot to metion, that especially as ambient temperatues run hotter, air cooling becomes much less efficient than water up to about 140 F and maybe even higher.

I've run a 3.2 GHz Intel P4 Prescott and the Northbridge for it and a 800 XTPE all in a single loop and gained only 4 C along the way. Of course, this is with the room at 78 F, not 100 F, so I've built a bit more fudge room into the actual heat gain he might encounter. I doubt that even 6 P4's of the type he's talking about would produce more heat at stock speeds and voltages than my setup did. So estimating 12 C for the entire loop is a pretty conservative estimate.

Of course again, this would all be very dependant upon having both the radiator with fans that's big enough and the pump to push it all along.