Water Cooling Kit Versus Air - Power Consumption - Because it's Earth Day

[ricoviq]

Anyone have any thoughts about the power consumption of water cooling versus air cooling? I've heard that the liquid cooling does in fact consume more power, but haven't seen the general comparisons. I have my build started and I'm about to pull the trigger on an IB CPU tomorrow. But I need help choosing a cooling solution.

My build video is Here

I'm going to OC slightly, but the comp will be running several hours a day and need to keep my power bill in mind too.

As you can see in my vid, I have everything except for my CPU, Heatsink, and GPU (still using it on my old build). I don't mind spending around $75 for a CPU cooling solution, but need it to be able to fit in my case and not interfere with a 7850 GPU and my UD5 MB and Vengence DIMM's

 

[natto fire]

From a pure physics standpoint, air cooling is more efficient. This is because we are surrounded by the atmosphere, where there are pumping losses in using a liquid coolant for heat exchange.

This is all YMMV overall though, because there are so many variables. In the end, the differences in power consumed to cool the same amount are probably negligible.

 

[ricoviq]

So basically, just chalk it all up to the computing Gods... It being earth day and all, figured I'd try to account for power consumption and all. Though I'm no liberal by any stretch, I still think about that stuff, mostly because I do get the power bill...

 

[jesh462]

You should be able to directly compare kwh between your selected pump (and accompanying fans), and what a HSF draws off of the power supply. Those are the only parts in each that use electricity.

I would assume watercooling would end up slightly more expensive power wise. You will generally have at least 2 120mm fans, plus the draw from the pump. It probably isn't a huge difference at all, but it's there.

 

[BrightCandle]

For equivalent cooling the watercooling would be +20W because of the pump power consumption. Other than that you could account for more slower fans produce less power consumption that less faster running fans so in theory there comes a point where watercooling uses less electricity. But in doing all this you haven't accounted for the material costs, and here watercooling looses a lot because it uses lots of copper and brass and combines it with quite a lot of flexible plastic.

 

[jesh462]

20w? That can be made up for by unplugging that night light in your restroom. You're totally right about the materials, though. A lot more goes into water cooling, and if you're thinking about it in a green way, that's more energy expended. Of course then you could look at where individual components are manufactured. Your block and pump may be made in California, where no coal is burned for electricity, and that HSF could be made by indonesian slave children heating up metals in a put over a tire-burning fire.

 

[ricoviq]

20w? That can be made up for by unplugging that night light in your restroom. You're totally right about the materials, though. A lot more goes into water cooling, and if you're thinking about it in a green way, that's more energy expended. Of course then you could look at where individual components are manufactured. Your block and pump may be made in California, where no coal is burned for electricity, and that HSF could be made by indonesian slave children heating up metals in a put over a tire-burning fire.

Your post had me rollin! LMFAO! But nah, motivation isn't to be green, just to save a few bucks a month. But if the difference is that small, I'll go with whatever is most effective for around $75.

 

[SZLiao214]

I don't see how water cooling could use less power since you are running fans and pumps. When you use a nice big heatsink the fans you use really don't have to be that strong.

 

[EarthwormJim]

Water cooling can potentially use less power if it is keeping parts like the GPU significantly cooler.

Transistors will use less power at the same switching speed if they are kept at a lower temperature.

 

[richaron]

Your post had me rollin! LMFAO! But nah, motivation isn't to be green, just to save a few bucks a month. But if the difference is that small, I'll go with whatever is most effective for around $75.

At that price there's no point going water. A nice air cooler will perform just as well & will use less power.

Water cooling can potentially use less power if it is keeping parts like the GPU significantly cooler. Transistors will use less power at the same switching speed if they are kept at a lower temperature.

I would be interested to see this tested. I think at the upper temps you're probly right, but a larger system will put much more strain on the pump (which is already drawing much more current than a fan).

 

[jesh462]

Water cooling may be beneficial in systems that spend a lot of time at load, like folding, etc. That's where water cooling really shines, not idle temps. It may be possible to design a loop that could outperform air (electricity consumption), but I think it would have to be specifically designed to do so.

I'm glad you got a laugh out of it. =) .
I'm not sure you can get water cooling together for 75 dollars. If you can, it may not be pretty. However you can use the same water cooling setup even after you change sockets, so you can keep that cost savings in mind when comparing water and air, in addition to the electricity.

 

[greenhawk]

given how much power a water cooling setup might use over just a single fan, I would not be surprised if overclocking your cpu would result in a higher power consumption for a given period of time.

 

[EarthwormJim]

I would be interested to see this tested. I think at the upper temps you're probly right, but a larger system will put much more strain on the pump (which is already drawing much more current than a fan).

I remember there was a very specific review that compared a water cooled GTX 480 to a stock cooled GTX 480 with power consumption data. I couldn't find it.


The best I could find was done only on an air cooled card, but the fan was manually controlled to target certain temperatures. Here's a link.

http://www.techpowerup.com/reviews/Zotac/GeForce_GTX_480_Amp_Edition/27.html

I would love to see if that linear trend continues below 80 degrees C.

 

[richaron]

I remember there was a very specific review that compared a water cooled GTX 480 to a stock cooled GTX 480 with power consumption data. I couldn't find it. The best I could find was done only on an air cooled card, but the fan was manually controlled to target certain temperatures. Here's a link. http://www.techpowerup.com/reviews/Z...dition/27.html I would love to see if that linear trend continues below 80 degrees C. EarthwormJim is online now Report Post

Cool thanks, I haven't seen much about that but it makes sense. They say it's gaining 1.2W per °C at the ~250W of the 480 (almost same as a oc'd CPU + GPU now). Assuming it's linear & a pump will draw ~10W; I would guess a water cooling kit would need to be almost 10°C cooler than air (@~250W) to start being more efficient overall.

Sound pretty easy. But it would need to be much more efficient under load to average out the ~10W idle penalty. And at $75 a crappy closed loop water setup is only for the cpu & can barely (if at all) outperform a nice air cooler.

 

[EarthwormJim]

Cool thanks, I haven't seen much about that but it makes sense. They say it's gaining 1.2W per °C at the ~250W of the 480 (almost same as a oc'd CPU + GPU now). Assuming it's linear & a pump will draw ~10W; I would guess a water cooling kit would need to be almost 10°C cooler than air (@~250W) to start being more efficient overall.

Sound pretty easy. But it would need to be much more efficient under load to average out the ~10W idle penalty. And at $75 a crappy closed loop water setup is only for the cpu & can barely (if at all) outperform a nice air cooler.

10 degrees is easily attainable. Most full coverage blocks can keep a GTX 480 in the 40 degree range. The air cooler on a GTX 480 can hardly keep it below 90 degrees. So that's a 50 degree difference.

The OP did say kit watercooling, but that makes this discussion boring since most kits perform worse than top air cooling.

 

[BrightCandle]

Hmmm interesting. The same thing happens on the CPU at load as well where you can often get into the 20's at idle and 40's at load compared to 30/40's and 60/70's on air. Likely reasonable power consumption reductions there as well. I doubt any of this "pays" for the cost of making the components to begin with but at least it helps.

 

[richaron]

Yes, there would be several changes Id make to my above example if we were going to generalise... Proper water systems will use much more than 10W (20-30W), & we don't know if there is a linear progression down to the ~40°C mark. But more importantly I was talking a single device @~250W.

Using the minimum ~10W water system & single processor ~150W I would guess you need closer to 20°C difference (over high end air) before you start seeing efficiency gains. Then you find average vs idle.

Edit: I'm just kinda extrapolating from a single set of numbers..... I'm more than happy go hide under a rock if people come here with actual results :sneaky:

 

[natto fire]

Hmmm interesting. The same thing happens on the CPU at load as well where you can often get into the 20's at idle and 40's at load compared to 30/40's and 60/70's on air. Likely reasonable power consumption reductions there as well. I doubt any of this "pays" for the cost of making the components to begin with but at least it helps.

Yes, the CPU will use less power but it is insignificant compared to the extra energy you expelled to get it that cool.

I don't know where you guys are finding water pumps that draw less than an amp at 12v. Even if these pumps exist for PC cooling, I can't imagine they are going to be that effective.

In this scenario where the CPU is running cooler and saving power, what is cooling the VRMs, that are going to be affected by temperature rise much more than the CPU die?

Convection happens in both systems, is essentially free, but happens more readily in air cooling. So I said it once and I'll say it again, based on my limited knowledge, air cooling is more efficient on a computer scale. Really the most efficient way is to passively cool your components, and if that can't be done with air, there is always oil immersion.

All of this hair splitting we are doing over the total electricity wasted is probably equivalent to leaving the fridge door open an extra minute or so, though.

 

[ricoviq]

Yes, the CPU will use less power but it is insignificant compared to the extra energy you expelled to get it that cool.

I don't know where you guys are finding water pumps that draw less than an amp at 12v. Even if these pumps exist for PC cooling, I can't imagine they are going to be that effective.

In this scenario where the CPU is running cooler and saving power, what is cooling the VRMs, that are going to be affected by temperature rise much more than the CPU die?

Convection happens in both systems, is essentially free, but happens more readily in air cooling. So I said it once and I'll say it again, based on my limited knowledge, air cooling is more efficient on a computer scale. Really the most efficient way is to passively cool your components, and if that can't be done with air, there is always oil immersion.

All of this hair splitting we are doing over the total electricity wasted is probably equivalent to leaving the fridge door open an extra minute or so, though.

I'm seeing the light here. I'll probably end up cooling with some really nice Heatsink and call it a day. Noctua has been popular among OC'ers not looking to get crazy.

 

[richaron]

Yes, the CPU will use less power but it is insignificant compared to the extra energy you expelled to get it that cool.

Obviously our "hair splitting" went right over your head...

But we agree on the OP should be using air.