jfrydom, on 15 January 2011 - 04:58 AM, said:
Miha i completely disagree with you on your statement about radiator thickness. Black ice GTX and Koolance radiators have a high fpi, which means they best work with high rpm fans. XSPC, Thermochill, and Feser radiators are good for lower fan speeds and silent applications. The range in which the low fpi radiators becomes worse then the high is from about 1300 rpm to 1800rpm. The koolance CU1020V which is about 35mm (iirc) is about as good as much thicker TA 120.3 at about 1500 rpm.
With that image i show the lack of validity in your "experience" a 580 has about a 250w heat load, and a HIGHLY OC'ed i7 is about 300watts. That is about an 18% increase in the heatload vs the graph, whereas 2x240 rads would be a 33% increase in cooling surface area. Miha a single loop will always provide the best UNIFORM liquid temps. If he really want to cool one component more than the other you could have 2 loops, but putting the gpu first should suffice. 2 loops can also be beneficial if you swap out gpus often as it allows you the ability to keep a part of the cooling system intact. 2 loops will be less efficient because as temp decreases cooling "potential" needed to bring it lower increases. I believe Miha's loop has problems, his 980x runs a t 75c or higher, it would be in your best interest to ignore any "personal experience" evidence he gives you with his current loop.
That's the difference I guess. A watercooling to me is low noise. What purpose does a watercooling have if you put 1200+ rpm fans on there. To most of my friends I'm nutts already for putting 1200 rpm fans on my radiators.
The idea of a water cooling to me is having a well cooled environment with 700-800 rpm fans, so it's silent too. My fans spin @ 700 rpm most the time... Wattage is a relative term for radiators since it depends on the fans you put on there.
But with same fans speeds, with a realistic low noise system @ 800 - 1200 rpm, a bigger radiator wins over a small one. I guess these are the differences between European views at things and American views. You guys seem to be way more tolerant to noise than we are. I build until now about 10 water cooling systems for various people, after having changed my own dozens of times. I think reading reviews is less valuable that if you try it yourself.
My CPU runs @ 78°C on one core if I fold or stress test. Otherwise I never ever see more than 70°C on any core. that puts the CPU temperature @ around 55°C and the water temperature around 28°C.
To go to the graph you put in. Only the blue with 600 rpm fans and red with 1000 rpm fans is really interesting. What's the purpose of having a watercooling if you're gonna putt fans the size and noise level of delta fans in there?
But maybe our friend from Singapore won't mind being forced to put headphones on while gaming....
Nofear, on 15 January 2011 - 04:09 AM, said:
That loop in the pic goes like so:
Res=>Pump=>Rad=>Gpu=>Cpu=>Mobo
Although, as long as you have Res=>Pump the rest of the layout doesn't matter (the pump should always be below the res)
That is a very weak setup, although it's true that the water temperature becomes more or less constant in a loop it's always best to put waterblocks first in the loop which give away less heat.
Also I always suggest you put the radiator at the end, because it slows down the water flow the most.
Res=>Pump=>MB=>CPU=>GPU=>Rad
Usually that leaves to the best result. The MB doesn't heat up the water much, the CPU gets there the maximum flow as the next, the GPU comes last because it usually heats the water the most. If the MB would come last the the water which got heated up by the CPU and GPU would increase the MB's temperature for a few Degrees.
You MB can get up to 60°C hot, your CPU up to 67°C (with the cores @ 81°C) and the GPU can get up to 90°C hot. That's why you put it into that matter. The difference is small but the weaker the pump and the lower the rpms on the fans the bigger difference it makes...