Originally Posted by num1son
I reviewed the H220 for ThinkComputers.org
and tested it fairly extensively(2 part review 2nd part coming out soon). Adding a second rad like the XSPC RX 240 is highly recommended if you want a GPU in the loop. Don't expect to be able to run a GPU in the loop with just the stock radiator, especially if you plan to overclock. There just isnt enough surface area to dissipate all that heat.
So you must be Derrick then, right?
I liked the approach you took in your review (the overclocker point of view), it departs from the mere enumeration of temps, and focuses on purpose. Well done.
But I am a bit surprised by the opinion you are expressing here above. As you may know, we ran TWO GTX 680's in SLI at CES while also cooling the Asus Maximus V Chipset, with just one radiator.
One hint as to what point of view you are coming from is in your "especially if you overclock" remark. And I will grant you that for extreme overclocking, a single dual 120mm radiator cooling CPU and GPU might not be the optimum setup, however this is strictly extreme overclocking.
So I do believe that your remark should be thoroughly qualified, and here are my comments in this respect.
For starters, the vast majority of users is not composed of extreme
Secondly, there is plenty enough surface area to cool a CPU and a GPU in the loop (even two GPU's for that matter), in an everyday use environment and especially with the type of CPU overclock that the majority of users will apply (the point and click type).
Here is why:
1/ Most applications only place load on either the CPU or the GPU, not on both. So when you are running a CPU intensive app, your GPU's are a t idle. Therefore all the cooling capacity of your rad is dedicated to CPU, and running GPU's in the loop will only cost a moderate rise in coolant temp due to the moderate GPU load (about 50 Watts per GPU at iddle or less depending on brand etc..).
2/ In the case of GPU intensive apps, Playing a game for example, it doesn't load your CPU. So when you are playing Crysis or what have you, the radiator is only busy cooling the GPU(s), and there's plenty of head room to do that at substantially lower temps that stock.
3/ Regarding item 2 above and with respect to graphics cooling, another reason is that despite their very high thermal load, GPU's have a larger die surface area than CPU's. It results that the heat flux is much less concentrated than that of CPU's, and that's the reason why you see such lower temps in your liquid cooled GPU's (running in the 50's) than what you see with CPU cores (running in the 70's or higher).
So in real life what happens? You are running a game with your SLI rig, and your GPU(s) run in the high 50's. Problem? Not that I'm aware so far. Meanwhile, what happens on the CPU side? Your coolant temp has been rising because of the GPU load. So your CPU iddle temp has been rising. Problem for overclock? Not so either, since there is no load on it.
Now, I know that many reviewers will try to place load on both CPU and GPU at the same time (running prime and Furmark at the same time for example), and I've done it myself for testing purposes, but let's say it clearly here: this is not representative of everyday use.
Yet, we can still ask the question here, what will practically happen in such cases? Very simple: coolant temperature will rise substantially and CPU and GPU's temps will also rise accordingly. You may see you CPU cores running in the high 80's, and you GPU temps running the high 60's. If you have an extreme overclock, system stability might be compromised at that overclock level. So if your goal in life is to run Prime and Furmark on your computer all the time, AND you only use that single H220 radiator, then I guess you'll need to lower your CPU overclock and voltage (for example tuning down from 4.8 to 4.6 like I did) and system stability will be restored.
In conclusion, I would certainly not tell people that you can't run a GPU in the loop without adding a rad. In fact I'd say exactly the opposite
: go ahead and add your liquid cooled graphics card to the loop, and see how it goes based on your own usage. I am willing to bet that in 90% of the cases, this will work out just fine. And for the 10% or so users who are dissatisfied with the results for whatever reason, I think there are more choices than just adding another rad, and these choices are based on the type of usage and personal goals/circumstances. At the one end of the spectrum, if budget is an issue and noise is not a grave concern, then you can try adding fans for push pull, or swap our fans for higher RPM and Static Pressure, and on the other end of the spectrum, say if budget is not an issue and noise IS a concern, then adding a radiator is indeed the route to take. The beauty of it all, is that the H220 allows you do choose, unlike most other AIO's out there.Edited by gabe@swiftech - 3/9/13 at 1:36pm