View Full Version : sli water cooled system
arctic-k20
04-18-2005, 06:42 AM
Well i finally took the time to take a couple of shots of my water cooled sli system. Yes I know things are dusty and yes I know things are messy but I havent had time to tidy things up yet...
My system is in my signature but here are stats:
2x MCP650's
2x 77' Bonneville Rads
2x Dangerden Manze4 Low Pro's
1x Swiftech 6002
4x Panaflow M1A's
The way things flow is as follows:
Pump1 > GPU1 > CPU > GPU2 > Pump2 > " Y " > RAD1 / RAD2 > " Y " > Pump1
All of the blocks are in series and the rads are in parallel. I had to go GPU to CPU to GPU because of how close the gpu's are on the dfi board - but the temp difference is negligable.
Idle temps are:
cpu: 28c
gpu1: 29c
gpu2: 29c
Load temps are:
cpu: 38~39c
gpu1: 34c (using everest)
gpu2: 34c (using everest)
:toast:
Bla$nA
04-18-2005, 07:16 AM
nice temps ;>
arctic-k20
04-18-2005, 07:21 AM
thanks!!
bimmer01
04-18-2005, 07:45 AM
any more closer pics?
MaxxxRacer
04-18-2005, 07:51 AM
dual weapon cores.. that thing is scary.
J-Mag
04-18-2005, 08:01 AM
If one were to cool just the GPU's would dual cores be necessary?
Also, did you see what the temps were if you only had one core?
MaxxxRacer
04-18-2005, 08:15 AM
only one core would be plenty for just gpus.
ferrari_freak
04-18-2005, 08:19 AM
nice setup and temps, closer and bigger pics would be nice tho :)
arctic-k20
04-18-2005, 08:19 AM
unfortunately i cant do that comparo because this setup went straight to dual heatercore water.......never booted up with air or one core so i dont have any results to compare it too...
closer shot....hmm let me dig around a bit...
arctic-k20
04-18-2005, 08:20 AM
also - the orginals are massive but the size is too large to upload...
arctic-k20
04-18-2005, 08:31 AM
ok i changed the pictures in the original post......they are 640x480 and i added 1 close up shot
J-Mag
04-18-2005, 09:38 AM
only one core would be plenty for just gpus.
Kinda what I was thinking...
How many ~Watts can each core dissapate?
Also, I can't remember, but what is the ~watt heat generation per Ultra? (cause the gt is at 1.4v...)
Artic: How are your cards faring with the extra .1v? I had heard some scary rumors that the PCI-E 6800GT's don't handle the extra voltage well compared to AGP version, b/c of the birdge chip. now I don't know if what I am hearing is BS or what cause to me it would seem that the PCI-E v should fare better than the AGP b/c the PCB is the same...
arctic-k20
04-18-2005, 09:51 AM
well the only time my gt's didnt hanlde the 0.1v was when the cpu/ram were not stable - an unstable oveclock makes the nvidia drivers extremely sensitive and they are usually first to give up...
but since the system is stable the cards have been fine.... (11,220 3dmark05) where as unstable i would score lower than 1 single card so it makes a huge difference...
as far heat generation im not sure??
MaxxxRacer
04-18-2005, 05:13 PM
J-mag you cant put a quantifiable number on the rads very easily. you have to be given a delta T of the water to air temp and a given flow... thats the only way...
arctic-k20
04-18-2005, 06:23 PM
i may switch to an independant loop for the cpu / gpu - at the time of install it was just easier to to add the second pump in series to the rest of the circuit...
pump>cpu>rad>pump
pump>gpu1>gpu2>pump
hmm - has me interested...
arctic-k20
04-19-2005, 01:00 PM
ok well i spent today completely rerouting the tubing for an independant dual loop setup.......PITA for sure in that small of a case...
the air is bleeding out while im at work.......hope to be able to get some temps tonight - not expecting much of a difference but worth a try...
and you thought the tubing was chaotic before - its very - ummm "busy" in teh case now ;)
J-Mag
04-19-2005, 01:05 PM
J-mag you cant put a quantifiable number on the rads very easily. you have to be given a delta T of the water to air temp and a given flow... thats the only way...
Yeah I gues syou are right... But i know alot of radiator manufacturers give a dissapation rating in Watts... They must use some standard (well I guess they could make up their own advantageous standard...)
vBulletin® v3.7.0, Copyright ©2000-2008, Jelsoft Enterprises Ltd.