Yes, I do plan to run the fans below 1000 rpm if possible. Perhaps as low as 600 rpm.
You make a good point about ST30 vs XT45. I think it may be worth the few less watts of heat dissipated (for a 360mm rad, 3W difference @10°C with fans at 600 RPM) to have a much less restrictive loop (0.23 PSI per rad instead of 0.48 PSI).
Yes, that's why I'd rather use a D5 like the MCP 655-PWM-Drive if I can get away with it, instead of the MCP-35X.
So you're saying if you take 2 pumps in series and dial them down to the same flow rate as a single pump, they will make less noise than the single pump? If so, then it sounds to me like I should go for 2x MCP 655-PWM-Drive. This would be quieter than two 35X's but should definitely provide sufficient pressure (compared to a single 655-PWM-Drive which might not). I think this might be the best option. At least, the best option for lowest noise... not the best for my wallet!
Oh and one thing: 617W???? No way.
1) All the consumed power is not translated into heat.
2) You can't dump all the heat in your loop. That's why some TIMs are better than others and why some waterblocks perform better than others. And you'll be using a Haswell CPU that has crappy TIM between the IHS and the die, that's even worse heat transfer.
3) 84W is the consumption including the iGPU. My 3570K consumes 90W at 4.5GHz, 1.33V. According to your formula, my 3570K should produce 77 * (4.5/3.5) * (1.33/1.2)^2 = 121W. That's 1/3 away from the real consumption.
4) And I'd said the same will be true for the GPU. It won't be able to generate almost 500W!!!
Your planned loop is overkill for a light CPU and a single GPU.
1) True. But it is just an estimate, and I wanted to be conservative.
2) True. But I am planning to use high quality TIMs for best heat transfer and I will likely be delidding the CPU if necessary after testing its thermal performance.
3) That's a very good point. So, I should lower my estimate a bit.
4) Actually, from following the r9 290x and 290 owners club thread, I've seen that it's quite possible to generate that kind of wattage with an aggressive overclock. These cards eat wattage like no other. I may not be able to achieve such a high overclock, but I used that in my estimate to be conservative.
Additionally, from the thread I linked, I calculated that my proposed loop could not actually dissipate all of that estimated heat with a 10°C delta. So, if that estimate is too high that would be a good thing
You may be right, my planned loop may very well be overkill. However, I will be aggressively overclocking two notoriously power-hungry components and I want to cool them as much as possible with the absolute minimum of noise. I totally get that it isn't necessary, but I find this kind of challenge fun
Edited by Sir Awesome III - 11/21/13 at 4:51pm