Sign in with
Sign up | Sign in
Your question

Had enough need to Watercool for High end

Last response: in Overclocking
Share
a b K Overclocking
July 4, 2012 8:50:46 PM

Hi everyone

My specs
Nzxt Switch 810
OCZ ZX 1000
I5 3570k
Gigabyte UD5H
8GB
2X EVGA 680 SLI

I have decided to go with watercool for my rig for obvious reasons too loud and too hot, my whole room literally. Now I know I need a 240 and a 360 rad. Could you guys give me some recommendation for all parts required?


Thanks

More about : watercool high end

a b K Overclocking
July 4, 2012 8:57:41 PM

Watercooling will NOT make your room any cooler. All water does is move the heat from your parts quickly out of the case and into your room. Noise will probably improve, yes, and your parts will get cooler. Just making sure you know what to expect. Most people go water more for the aesthetics and for the fun of putting it together than for the performance.
m
0
l
a b K Overclocking
July 4, 2012 9:24:56 PM

kajabla said:
Watercooling will NOT make your room any cooler. All water does is move the heat from your parts quickly out of the case and into your room. Noise will probably improve, yes, and your parts will get cooler. Just making sure you know what to expect. Most people go water more for the aesthetics and for the fun of putting it together than for the performance.



Thats is true but when I am gaming I got two jet engines spewing out 75 degree temps. No doubt that I really want those wonderful WC aesthetics as well
m
0
l
Related resources
a c 324 K Overclocking
July 5, 2012 1:54:20 AM

Specific heat and thermal conductivity of water; these are the terms that discuss how watercooling is much more effective at cooling than air alone.

The new sticky will have a section with an overview.
m
0
l
a b K Overclocking
July 5, 2012 4:30:04 AM

redeemer said:
Thats is true but when I am gaming I got two jet engines spewing out 75 degree temps. No doubt that I really want those wonderful WC aesthetics as well

Those chips will generate exactly the same amount of thermal energy with water. If you want the aesthetics, go for it.
m
0
l
a c 324 K Overclocking
July 5, 2012 2:09:21 PM

Quote:
Those chips will generate exactly the same amount of thermal energy with water. If you want the aesthetics, go for it.


Water has the ability to absorb heat energy far faster than air, which is why you see core temperatures of a GPU drop from 75C to 40C at load. Water also has the ability to absorb tremendous amounts of heat energy before the overall temperature of the water raises a single degree C. Couple these concepts and the reverse ability of water being able to dissipate heat extremely well through a radiator, as well as a radiator having far more surface area than a normal air cooler, you are dissipating less heat over the entire course of the day. So no...it's a bit misleading that this is a 1:1 comparison, when it's really apples to oranges. The heat from your power supply at load will still be the same since power consumption would be equivalent...even if you factor out the fans on CPU/GPU coolers and factor in 20w of pump draw and the handful of watts for radiator fans.
m
0
l
a b K Overclocking
July 6, 2012 1:38:06 AM

You're missing the basic physics here. The chip, under load, releases a certain amount of heat energy, whether it's into air or water. No matter whether it's dissipated slowly as the water cools down to room temperature or quickly by a fan, it's going to get into the room. The chip is cooler, but the heat is still there, still in the room. You're dissipating exactly the same amount of heat over the course of the day.
m
0
l
a c 324 K Overclocking
July 6, 2012 2:56:10 AM

I'm not disagreeing that you are still producing the same amount of heat in watts, the main point I'm debating is that the overall efficiency of water vs. a straight air cooler and how it's being dissipated. I am also saying that once you shut down your water loop, your rads are still releasing a large amount of heat that has been stored within the loop throughout the cycle due to the high specific heat of the total volume of water. While it absorbs and slow dissipates heat via the radiators, it always operates at a level of thermal equilibrium based on the delta of the loop. This directly correlates to a specific amount of thermal energy being maintained at the operating temperatures of the loop.

However, due to running cooler, hardware also runs more efficiently lowering the amount of power needed to operate in the first place; flowing electrons face less resistance at lower temps, therefore draw less power.
m
0
l
a b K Overclocking
July 6, 2012 3:49:52 AM

One of us is getting a little tangled up in your terms there, and I'm not sure which of us it is. Stick to plain English. I'm pretty decent at physics, but I'm not getting your point.

I'll be surprised if that last bit about higher efficiency at lower temperatures is true. Googling doesn't pull up anything immediately apparent about it. Can you give a link on the topic?
m
0
l
a c 324 K Overclocking
July 6, 2012 1:56:24 PM

No links that I can think of...mostly what would appear to be common principle. I'm not a physics guru, so you definitely have the upper hand when it comes down to it. Perhaps I'm wrong with my assumptions and I'm more than willing to admit that you do have a very solid debate. At the end of the day, I do agree that the same CPU does put out the same amount of heat energy, regardless of how it's cooled and dissipated- it's the efficiency and process by which it's dissipated that was the basis of my argument.
m
0
l
a b K Overclocking
July 6, 2012 8:19:36 PM

It's possible that your point about the high specific heat of water may make a useful difference, as long as the rig isn't run at full blast for hours. Once the water gets up to its maximum temperature, the temperature at which equilibrium is reached between the heat entering it from the processor and the heat exiting it from the radiator, it'll be doing exactly the same thing as the air, just moving the heat over to the radiator before releasing it. Before that point, though, the heat will be stored in the loop and reservoir, keeping the room temporarily cooler than air might.

With an absence of evidence to the contrary, I'm going to fall back on my previous belief that electronics do not use a significantly smaller amount of power at lower temperatures. I don't know enough about the architecture of CPUs (all those tiny wires are a mystery to me) to do any sample calculations, though. How much resistance do all those tiny connections have? How much current passes through each of them? I'll ask my computer science teacher if he knows anything about it.
(Email's now out!)
m
0
l
a c 324 K Overclocking
July 6, 2012 8:44:43 PM

The specific heat issue is somewhat a unicorn since we rarely run our complete system at 100% (nor would we really want to). The idea is to be able to adequately cool for a worst case scenario of 100% TDP pegged to the redline performance, but it isn't the case in real life much of the time. This is why we often calculate for TDP over limits of what is needed and always coast on over-radding our loops.

Agreed- loops will always reach a working equilibrium based on delta and will dissipate/absorb much like busy club on a Saturday night and a 'one in, one out' policy. This was a great discussion...you should definitely pop in more often.
m
0
l
a b K Overclocking
July 6, 2012 8:48:01 PM

I love when people are reasonable on the internet. I'll get back when I get a response from the teacher.
m
0
l
a b K Overclocking
July 7, 2012 12:41:21 AM

Did you read the sticky watercooling guide yet? Read it and reread it until you understand everything in there before planning on how to set up a water cooling loop with radiators.

Watercooling is expensive, especially when you add 2 GPU into the loop. However, the temperature of the CPU and your 2 GPU will be substantially lower if done properly. A lot of people think watercooling is just for look, but its a whole lot more than that.

Watercooling will give you temperature that no air coolers can achieve. In addition to that low temperature, noise is also reduced substantially. In order to achieve this, you need to understand how much heat is being generated (in terms of estimated total TDP in Watts) of your components (CPU + 2 GPU) and choose the appropriate radiator/pump/CPU and GPU waterblock. All this knowledge is in the sticky. I was just like you couple weeks ago when I first stated my watercooling rig but I read the sticky thread and now I have a pretty good knowledge of watercooling.

My watercooling rig is also very similar to yours btw, got same case and same GPU/CPU.

m
0
l
a c 324 K Overclocking
July 7, 2012 3:25:19 AM

BTW, I just updated the sticky today with the new version with more info and more space for expansion.
m
0
l
a b K Overclocking
July 7, 2012 3:05:58 PM

Got a response. He forwarded it to an alumnus, an electrical engineering student at the Cooper Union.
Here's what I have, all from the email sent by the alum:
Quote:
Disclaimer: I do not have any numbers to back this up, nor have I ever built a water cooled pc (not worth the cost and hassle for me).

It is certainly true that higher temperature increases resistance and transistor switching time, but the effect is relatively minor (though not insignificant) in the typical computer operating range, say, 30-80C. Beyond 80C, CPUs start to become unstable because the resistance is too high or transistor switching time is too slow. This instability can be mitigated by increasing the voltage, but power is proportional to voltage squared, so you're using much more power and generating even more heat.

Assuming we maintain a constant voltage (and are in the stable temperature range), a chip at 80C would certainty use more power than a chip at 30C, but the difference would be relatively little. The power difference between running the pumps and fans (because you still have to cool the heatsink your fluid is running through) compared to just fans for air cooling would probably be (at least) an order of magnitude more than the power difference from the resistances. On top of that, the initial cost of a good (read: not bottom of the barrel, semi-likely to leak) water cooling setup is significant.

In short, the power savings of running at a lower temp are insignificant compared to the additional power used by the water cooling system, and the even power cost of the water cooling system is relatively small compared to the hardware cost of the water cooling system. As such, from a cost point of view, it absolutely does not make sense to use water cooling on a desktop. It makes even less sense on servers for the same reason, except you usually have tens to tens of thousands of servers at a facility, and if something doesn't make sense for one machine and the cost scales linearly (the capacity of your water cooling system has to scale linearly with your thermal load, there is no n-for-one deal), it won't make sense for more machines.

Water cooling is never cost effective. The closest thing to cost-effective water cooling is building a server farm next to a river (or other body of water) and using the already cold water for building-wide cooling. Water cooling only makes sense where it is required for technical reasons (high-end overclocking, supercomputers which would overheat otherwise, etc.).

If you want, I could forward this question to a friend of mine (also EE at Cooper) who builds water cooled gaming rigs, see what his thoughts on this are.
m
0
l
a c 324 K Overclocking
July 8, 2012 2:37:40 AM

That's a lot of good info, and I agree that watercooling isn't something for the mainstream or mission critical systems, even though watercooling pumps and setups of good quality have a long MTBF. Yes, cost is prohibitive in most instances so that's definitely an accurate statement. Watercooling is much more a hobbyist cooling solution, but it does perform well, and most common pumps pull less than 25-30w of power in use, but still would be less than the wattage difference between an identical hot/cool processor.
m
0
l
!