How many is too many?

[Nanometer]

At what point does adding more radiator power start to get pointless? Obviously if the water is 20 degrees above ambient you still have some way to go. But is 2 Quad radiators overkill for a single CPU? I just finished heat transfer at my college, and I might be able to figure out the gains, but I was curious if any of you guys have had some first hand experiences with it already?

[TJ TRICHEESE]

definitely overkill, i have 3 285's on one quad and they dont go over ~50-55c all on full load and the water temp is about ~5-10c above ambient.

[Church]

Too many rads usually is preceded long way before reaching it with other problems it brings along - like limited space of case for internal mount, too big weight/size issues even with external rad boxes, too much funds spent on LC hardware with almost no gains at some point, where you'd be better off going cascade cooling and so on..

[jcool]

With a Quad 480 Rad for one CPU, you are probably looking at water temps of 2 - 4C above ambient, depending on CPU Overclock, load, and fans used. With a 2nd 480 you'll maybe get 1,5-2,5C above ambient - not really worth it.

[tool_462]

http://skinneelabs.com/assets/images/Radiators/TripleV2/triples-v2_OC920_deltaT.jpg

[Hannibal Lecter]

The great value of having an overkill number of rads is that as you add rads you can decrease the speed of your fans. Thus you get a very high performance cooling set up with very low noise.

If you don't care about noise, save your money. The difference in cooling performance will not help your cpu o/c and even less your gpus.

[Church]

After some point decreasing fan speed gets pointless, as for best of quiet fans 1000rpm can be as silent as 600rpm. Now going fully passive liquid cooling route needs massive rad area increase (IIRC even very slow cooling with fans is 4-10 times more effective then passive)

[Vapor]

I see noticeable gains when going between one triple and dual triples (push/pull 900RPM Yates)....but 'noticeable' is really just anything greater than ~.75C in my book. (Typically, there's about a 3C difference between one rad and two rads for me...this is a CPU-only loop, but the CPU is overvolted way past ideal for the sole purpose of making more heat)

Going to dual quad will afford you a small drop in temps at the same noise level, but it's not without its drawbacks. Cost, weight, and the fact you need 8x120mm intake accommodations are probably the biggest drawbacks. If you don't have 8x120mm intake accommodations, there's a big drop off in efficacy of the second rad...so much so that I'd probably just avoid it if it were my system (in this scenario).

[Nanometer]

I'm going for completely silent loads, since I don't have a system that automatically powers up the fans at a certain temperature. I'm loading right now on average 66-68C with my current water cooling setup. I'm adding a third Swiftech 4-120 radiator. More for curiosity, since space isn't an issue and nor is the weight; I'm just hoping it doesn't fall over under its own weight!
I'm expecting to see about 3 degree decrease in load temps, perhaps more. I'll post results here if anyone is interested. later this week.

[zsamz_]

i got a 320 and a 220 cooling my i7 no big difference in 2 rads
the big difference is when i put 220 straight on a/c duct

[Nanometer]

haha wish I could have a chilled water system...

[Utnorris]

You can make one out of a window AC fairly cheap and simple. The cost comes with a controller if you are trying to stay above the dew point to prevent condensation, otherwise you have to insulate your blocks. I have a controller on mine so I can set the fluid temp to whatever I want. The controller itself cost $60, but the whole thing was less than $150.

[Serpentarius]

I'm going for completely silent loads, since I don't have a system that automatically powers up the fans at a certain temperature. I'm loading right now on average 66-68C with my current water cooling setup. I'm adding a third Swiftech 4-120 radiator. More for curiosity, since space isn't an issue and nor is the weight; I'm just hoping it doesn't fall over under its own weight!
I'm expecting to see about 3 degree decrease in load temps, perhaps more. I'll post results here if anyone is interested. later this week.

i'm interested :D

i've been thinking ... how far can i7 actually oc .. most the temps i seen here is above 60+ .. i wonder if with enough rads & pump ... dropping the temp by another 10c, will it helps?

[Nanometer]

Right now with my current setup I can get 4.4 GHz on 1.38. I'm still lowering the voltage because I am trying out some new settings. I expect it to sit around 1.36 with this particular chip.

One of my big issues, is that the CPU is nearly 30 degrees hotter then the fluid temperature. I'll load the CPU, and then when I end the program it shoots back down to 45c from 73c. It could be a few things, but I have no idea what thing..

[the finisher]

I'm always interested in overkill.:up:

[Vinas]

It's all about surface area... to a point. For instance, you need a good pump, good CPU block, good TIM, for it to make sense to continually add radiators. Since we know it's impossible to get below ambient with just regualr liquid cooling, it's important to pay attention to the quality of all the other components in your loop. I think this is why we're all so obsessed with reviews of the best components available.

So how much is too much? For me, it's probably at the point where I don't see at least a 2c load decrease when adding new stuff. When I went to a 480mm magicool from a single dd 240gts, I saw a decrease in load temps. Granted they are different rads with their own characteristics, but if adding another 240mm of rads got me 4c, shouldn't adding another 240mm of rads get me another 2c more? Before buying more rads, maybe get a temperature of your water now. Find out your delta between inlet and the water temp coming out of your last rad. Then compare those to ambient. Use the chart above to guess where another rad might put you at. Or better yet, calculate it out! ;)

[justin.kerr]

I have 4 triples and one quad radiator, and am kicking myself for not going with 5 quads.. won't make that mistake next time.
Fans are silent at 5V, but temps get too high, especially the GPU's.. at 7V the fans are noticeable, but temps are decent, and at 12V fans are loud as hell, but temps are fine.

[Church]

4 tripples, one quad and temps too high? There must be something really really wrong imho, like placing case near room heater rad or badly planned out airflow in case (if rads mounted internally).

[Razrback16]

You can make one out of a window AC fairly cheap and simple. The cost comes with a controller if you are trying to stay above the dew point to prevent condensation, otherwise you have to insulate your blocks. I have a controller on mine so I can set the fluid temp to whatever I want. The controller itself cost $60, but the whole thing was less than $150.

How much power does that eat up?

[the finisher]

I have 4 triples and one quad radiator, and am kicking myself for not going with 5 quads.. won't make that mistake next time.
Fans are silent at 5V, but temps get too high, especially the GPU's.. at 7V the fans are noticeable, but temps are decent, and at 12V fans are loud as hell, but temps are fine.

Yes, Overkill is good;) If you got the space, and the money, why not.:yepp:

[Nanometer]

This is what I have right now. I don't think 4.6 can be made stable without chilled water or single cascade.
http://img.photobucket.com/albums/v718/NanoMeter/45.jpg

[justin.kerr]

4 tripples, one quad and temps too high? There must be something really really wrong imho, like placing case near room heater rad or badly planned out airflow in case (if rads mounted internally).

nothing wrong with the setup, it just does not dissipate enough heat.
With Kombuster, extreme burn mode the 3 5870's overclocked to 1062Mhz they will hit over 40C.. that is way to high for my taste.

[justin.kerr]

with HT off should be no problem, here is one of my rigs with HT off.. runs cool as a cucumber.


http://i18.photobucket.com/albums/b140/hoss3430/prime-2.jpg

HT on, same speed, gained about 12-13C.. lol

[mk-ultra]

since when an overclocked gpu @ 40+ is too high?
did i fell asleep?

[justin.kerr]

in my world. :D 20's is good, 30's is pushing it, 40's way to hot.

[JoeBar]

It all depends on what everyone's used to... ;)

For me something is wrong in this thread however. Talking about cpu/gpu or whatever temps and not water temps is pretty pointless and misleading...

[mk-ultra]

in the REAL world gpu can live all his life @ 80+ but that's real world... :D

btw any chance of a picture resize to stop the side scrolling? laptop is not wide enough

[Nanometer]

19c ambient with fans on slowest setting. This is about 10 mins on prime, the temperatures had not changed, so I felt this was about the peak. High fans remove another few degrees C. Still testing, and all cores raised by about a degree and is definitely stabilized. A lot of water means it takes a longgg time to reach steady state. The build is very cumbersome to move, but cooling performance is pretty impressive.

http://img.photobucket.com/albums/v718/NanoMeter/19clowfan.jpg

[Nanometer]

http://img.photobucket.com/albums/v718/NanoMeter/0527002219c.jpg
http://img.photobucket.com/albums/v718/NanoMeter/0527002219b.jpg
http://img.photobucket.com/albums/v718/NanoMeter/0527002219.jpg
http://img.photobucket.com/albums/v718/NanoMeter/0527002059.jpg


Load temperatures have improved by 5 c on the hottest core, and 4 degrees on the coldest core.

[Waterlogged]

Good grief Nano! How is it that that Qmicra isn't doing wheelies? :eek:

[Nanometer]

It's close believe me lol

[Nanometer]

http://img.photobucket.com/albums/v718/NanoMeter/ee.jpg

supposed to be mcr220 + 320 and then adding the mcr420... i was trying to do it too quickly haha

1.38 Volts at 4.4Ghz Ambient air at 19c