SLI Blocks in parallel arrangement??

F

FatboyHK

Limp Gawd
Joined
Nov 14, 2006
Messages
268
Stealth-sli-parallel.jpg


This photo, which comes with the Swiftech Stealth's annoucment, caught my interest.... it will make tube management much easier... but performance-wise is it actually a good way to go?

In Swifech's annoucment it also mentioned that this parallel arrangement can also improve flow rate of the CPU block.... seem it is really a good idea, but why don't we see many people do it that way?

what I am thinking is something like this, for my EK 8800 Blocks:

parallel.png


if it is possible then the interior of my rig will be alot cleaner.
 
I would like to see how the first GPU is affected by the "T", when it comes to cooling.
 
Unfortunately the pic is low resolution so I couldn't see what was really going on. However, putting really hot GPU's in parallel has its advantages even if it cuts flow. If you have two hot running GPU's in serial then you heat up the second one from the first one. When you run in parallel both GPU's get the same temp aimed at them.

Since a drop in flow of 50% might only cause a 1-2 C degree difference you may come out on the positive side if the first GPU dumps 3-4 C degrees in the second one when run in serial.
 
so can I connect my EK block in the way I suggested? seem the primary block will have NO flow, but the law of fluid dynamics say water will flow from high pressure to low pressure, so the flow for both block will be the same, because they have the same water resistance. Or am I correct?
 
That's worse than serial. Try running the input to a Y that goes into both GPU's. The exit of both blocks goes into a Y that goes to the next place in your loop.
 
has anyone tried to do something similar before? if not let me experiment with it tonight :)
 
It won't work. Especially not with the EK blocks:

wbju9.png


I think it's pretty obvious.
 
TN : Do you have any idea how much heat a GPU would have to give to off to heat up the water 3-4C at any reasonable flow rate?

Even if we assume a horrendous 1LPM flow rate, to raise that water 1C the GPU would have to give off a bit over 100 Watts. Therefore you would need a GPU that gives off 300-400 Watts to raise the water temp 3-4C. Now please tell me what your low flow setup flow rate is? I bet it is higher than even this 1LPM. So if you step the flow rate up to 3LPM you need a GPU that gives off 900-1200 Watts.

See where I am getting at?
 
I have done an experiment, hmm, surprise surprise....

First, see my setup:

an EK res and 2 EK blocks, all in my bathroom
PICT0053.jpg


fill the res with a running tap
PICT0054.jpg


Water run out from both blocks
PICT0055.jpg


Use a bottle to measure the flow
PICT0056.jpg


both blocks need 15 seconds to fill a 500ml bottle. Done the experiment 10 times and the result is the same.
 
Nice experiment. However I have to point out that the way that is setup does not necessarily correlate to the proposed setup. I can't begin to tell you what effect connecting the tubing on the return side of the blocks will do, but it could change the way the water behaves. Just pointing it out.
 
FatboyHK said:
Erasmus354 do you have any idea to improve the experiment setup?
Click to expand...

There isn't really much you can do different without flow meters, aside from actually doing a loop both ways and see if there is a difference in cooling.
 
This can really cut down on tubing runs if this indeed work. I am in the middle of redoing my box so this is really good timing. I have my GTX blocks out, so I might give a trial run to see what results are with the DD. If the flow is the same for both blocks, I might give a real run to see what the temps are. My temps was about 55°C and 48°C idle for the GTXs and CPU about 40°C. Pump-->DoulbleRAD-->TripleRAD--GTX1-->GTX2-->CPU-->RES-->Pump.

Hopefully, two loops will help.
My goal is to have the temp on the GTXs lowered to the low 40°C with little as possible tubing runs and of course, quiet as possible.
 
I wonder, which all those different blocks, how can you balance coolant flow??
 
You must log in or register to reply here.
Back
Top