Quote:
Originally Posted by Petrol
Thanks for posting up some numbers, it's great to have something tangible to work with. First off, I take issue with your claim of 40% improved cooling since 29.9 degrees is about 95% of 31.6 and 72.5 is about 97% of 75. That's at most an improvement of 5%.
What delta air to water does indicate is a 40% improvement in overall rad performance, but considering you added 66.6% more rad to get that, a lot of potential is being lost.
Now, the fact remains rad efficiency is not the only variable in the big picture. Coolant flow is also a key variable, and the efficiency of the waterblock depends more on that than it does on rad efficiency. The point I have been trying to get at here is that Behemoth77 should try swapping to one rad and see how well it works for him. If removing that extra restriction from his loop allows for better flow, that could very well add up to the same 5% improvement that opt33 noted by adding a rad, but we won't know that until it's been tried because we don't have all the variables here necessary to make accurate calculations to assess the outcome.

Sigh, I may need to just apologize for blaming Shazza on all my posts missing....plus of xtreme is less repeating, the minus of xtreme is it is difficult to follow threads with half the posts missing...
But to answer your point, first need to agree on a couple things.
ambient = starting point = 24C.
delta air to water = dependent on fans/rad, ie that is the cooling component, my delta with 1 360 rad was 7.6C. Adding more rads/fans can only improve this 7.6C, hence calculating their effect on cooling efficiency by definition is only applied to this delta of 7.6C.
delta water to cpu = dependent on cpu (leakage, IHS die attach, settings, etc) + waterblock and tim/mount you use. = 75C7.6C24C = 43.4.
If you go back and read my post, I said ~40% improvement in delta air to water. Since by increasing/rad fans, it is only the delta air to water that is affected, calculating such efficiency is determined by ONLY the change in delta air to water. It is an error to apply rad changes to entire temp or anything OTHER than what it affects. 7.6C (360 rad)4.9C(360+280) = 2.7C lower, 2.7/7.6 = 35.5%. So I was expecting roughly little over 40% improvement (that 40% explained below), only got 35.5%. (NOTE I used 120 fans with zip ties, so my 280 rad functioned more like 240) But not even that 5% difference is from rad efficiency, most of that is the lian li filters on 280 impeded air flow ~1520%, the metal filters on 360 rad impede air flow ~10%, also blowing air in case adds another 20% in restriction,
which may be uneven to some degree. And yes their may be a 12% efficiency loss, but it is too small to measure as it gets lost in other variables error.
Also "adding 66% more rad" as you say. Here is easy way to look at it.
If my delta air to water = 8C with 1 360 rad, then adding a second 360 rad (fan speed/filter/etc all being same), you would expect to cut that delta in half, ie 50% improvement in cooling to 4C, by adding "100% more rad", IF IGNORING changes in rad efficiency, other errors. In fact what you end up doing, is very close to that. If some of that error is rad efficiency, dont know, just too small for me to separate from other errors/variables.
In my example, delta was 7.6 with 360 rad. If adding 240 rad (mine was 280, but used 120 mm fans with zip ties to use same fans, so for arguments sake will call it 240 rad, not to mention have done same test on my last setup with swiftech 240 rad)...but since added 2 of now 5 total 120 rads, one would expect a 40% decline in temps by increasing the cooling area, maybe 42% since 140 rad, or maybe only 38% if adding in fact filters are little more restrictive on 120 rad.
I have done this in past with exact same filters, exact same fans, and with and without filters, posted on Xtreme long ago under RGE back when I had a 360 rad and a 240 rad, hence I know my metal filters cause ~10% hit in performance, and lian li filters ~?17% cant remember, but were little higher. And when I should have measured exactly 40% decrease in air to water delta temps, it was only about 37%, ? how much was my error, and ? how much increase in efficiency of rads, I just know at most rad efficiency is very small %.
To really measure increase in efficiency, you need higher wattage to increase the difference, to separate it from error. My guess is Martin has probably measured it before, but I had no interest.
As an aside, with cpu loaded and gpu idle, by having both rads in front of my cpu vs gpu in front of cpu, I get 0.2 C better temps. With requires 34 runs of doing both, just to be sure.
Edited by opt33  6/17/11 at 10:42am