radiator calc

As i mentioned in my previous post, a quality thick 120.4 will be fine for 3x 580's, 3x 580's ain't that much more power hungry then a pair of em and an overclocked cpu in the loop and i managed to keep my 580's under 50c in summer benching with a pair of 120.2 rads.
 
Skythe GT's (AP-15) are great rad fans, i run mine at around 700rpm for general use and gaming and only crank them up to maybe 1400rpm for benching and they are still pretty quiet.
 
The source I used suggests 0.11 for 1000rpm fans, 0.05 for 3000rpm. So I imagine the 1200rpm ones you have in mind will be around 0.1. Close enough anyway.

Which 1200rpm fans are you getting? The CFM figure is into open air, so means little on a radiator. The noise spec on cpu fans isn't standardised, so could be recorded at 10cm or at 20m, i.e. it doesn't mean anything.

Unfortunately fan selection for a radiator depends on static pressure, where 38mm thick fans are significantly better than 25mm thick ones, and the pressure scales roughly as the square of angular speed. Thicker, faster fans are louder, and finding the right compromise between fan speed / water temp / number of radiators is (in my opinion) the hardest part of watercooling computers.

edit: Parallel does bad things for flow rate across the parallel section (and so does good things for flow rate in the series part of the loop). It's more complicated, conceptually at least, than putting everything in series. GPU's with full cover blocks lend themselves particularly well to parallel arrangement, matched radiators could benefit too.
 
Last edited:
Andrew Moore said:
Just to add to the above..

The cards can manage 1050 core, with the cpu at 5.3ghz for benching.

During this period the temps hit around 54c loaded on the GPU and the CPU after a good hour or so of messing around.

These temps soon drop down to around 25C-30C idle.

The rads in question are not any where near as high a capacity as an SR1 for instance and with that in mind I should think a tri sli 580 set up on any decent 480 rad would be more than sufficient.
Click to expand...

thanks! food for thought
 
JonJ678 said:
The source I used suggests 0.11 for 1000rpm fans, 0.05 for 3000rpm. So I imagine the 1200rpm ones you have in mind will be around 0.1. Close enough anyway.
Click to expand...

actually, I've just spotted these - these seem higher CFM / pressure and also quieter although yes point taken on the test not being standard

http://www.overclockers.co.uk/showproduct.php?prodid=FG-010-CM&groupid=701&catid=57&subcat=4

but higher pressure at the same RPM seems like a good thing

JonJ678 said:
edit: Parallel does bad things for flow rate across the parallel section (and so does good things for flow rate in the series part of the loop). It's more complicated, conceptually at least, than putting everything in series. GPU's with full cover blocks lend themselves particularly well to parallel arrangement, matched radiators could benefit too.
Click to expand...

I definitely wouldn't want multiple rads, but splitting 3 ways just before (and coming back together just after) the GPU's wouldn't be that difficult
 
Last edited:
rjkoneill said:
i would imagine that your tri sli 580s that run cooler than GTX480s would be fine on a 480 as long as you have a good pump.
Click to expand...

the nvidia quoted TDP is only about 6W less, I thought they improved the cooler quite a lot between the 2 as well and that was the main difference to why the 580's ran cooler?
in terms of watercooling them they should be about the same
 
andybird123 said:
the nvidia quoted TDP is only about 6W less, I thought they improved the cooler quite a lot between the 2 as well and that was the main difference to why the 580's ran cooler?
in terms of watercooling them they should be about the same
Click to expand...

no, 580's are a LOT cooler than 480's!
Had a 480 and a 580, same cooler, same brand and the 480 was considerably hotter :)
 
Krony said:
120.4 SR-1/XSPC RX/Thermochill TA would be my choice for tri-sli 580's or sli 580's and a cpu.

I was cooling an i7 950 and sli 580's with a pair of 120.2 XSPC RX rads and it was fine, i then upgraded my case and added a 120.4 SR-1 into the loop just because i had the space and it didn't really make much difference at all, maybe a few degrees at load and faster temp drop from load to idle but that's it.
Click to expand...

Im confused - isn't 2 x 120.2 the same as a 120.4?
A 120.4 is a long rad that fits 4 120mm fans, right?
 
edgedemon said:
Im confused - isn't 2 x 120.2 the same as a 120.4?
A 120.4 is a long rad that fits 4 120mm fans, right?
Click to expand...

Yes, what i was saying was i was using a pair of 120.2 rads which is basically the same as a 120.4 rad and it was fine benching at some decent clocks, i didn't swop them for a 120.4 i added a 120.4 so basically the equivalent of a 120.8 or 240.4 :P

I bought an LD Cooling PC-V8 case and had the space to add in another rad so i did just for the fun of it and as i mentioned before it made hardly any difference which means that with what i was cooling which will be fairly similar to what the OP is wanting to cool the pair of 120.2 rads were plenty.

If u click my sig which needed updating months ago i know u can see the old setup in the Antec 1200 and the new setup in the LittleDevil which made hardly any difference to temps at all, i was considering adding a 3rd 580 once my system started to struggle and the extra rad would have given me the headroom for that but as it goes there is nothing that it struggles with atm to bother, i'll see how Kepler performs first and then decide on a 3rd or to sell em and get a pair of 780's, i was looking at upgrading a while ago to a U3011 which might need a bit more horsepower in some games but atm that is on the backburner.
 
Last edited:
I think 962MHz @ 1.15v was my highest stable run with the cpu @ 4.25GHz @ 1.38v, temps on the gpu's while benching normally peaked at about 46c, i'll try and dig though my screens.
 
You must log in or register to reply here.
Back
Top Bottom