• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Y's and SLI

catkicker

catkicker

Limp Gawd
Joined
Mar 30, 2005
Messages
388
I've been doing watercooling now for awhile. Seems like I'm always setting up something different. I'm getting ready to setup dual 7800GT's with DD Acetal MAZE4 blocks.
I know the basic config for running the tubing in one and out to the other and then back to rad

What I was wondering was if this would work well


This would be 3/8's " tubing. the tubing length from the Y's to the blocks would be the exact same length, so water flow to and from blocks should be equal.

I did try search function on this topic and did not fins anything.
 
Look up parallel and serial loops. I've read something about this b4 and am also considering the same setup with my SLI next week. I've heard that performance is not as good when the flow is split up equalll before entering the gpu blocks. Check out extremesystems.org as I believe I saw it there. Please post anything you find out. I would like to know. You research, me benefit
 
I just finished a setup like this last night, I have 2 7900gt cards with 2 koolance blocks (the 1/4" ones), a swiftech apogee, 2x120mm radiator, swiftech MCP655 pump. Initially i just had 1 block going into the other and flow was so restricted my temps were suffering. I picked up a 3/8" Y splitter and streched the 1/4" tubing over it, it flows 3x better now.

While the system was working the tiny little bubbles out i could see the flow through the koolance blocks, it looked pretty equal as far as how fast the flow was goin through each block.

When I get it hooked back up tonight or tomorrow I'll let you know how it worked out.
 
Here are some pics, I dont have the 2nd 7900gt set up because its in the rma process. Hopefully I will have it back next week.


IMG_0014.jpg


The CPU was bouncing between 42c-43c while I was running dual prime95's, and the video card idles at 39c, I haven't fired up 3dmark yet to heat it up, i will do that next.

I would really like to get the 1/2" Danger Den Koosah SLI blocks, but I just cant spend $259 on them after I just spent $300 on WC gear.
 
@opty170:
Looks like you should be able to reduce your tubing length significantly using those fittings.
(apologize in advance for 3rd grade level Paint skills....)
yloop.jpg

BTW, how difficult was it to fill the rez with it placed like that?
 
seems to me (and im a liquid cooling nub) that by doing your parallel liquid cooling you could get significantly better temps. being that GPU 1 temps have no (or little) effect on GPU 2 temps.

the reason could is underlined is because you are now decreasing the water pressure by 50% (obviosly), and as such if the liquid in the block is reaching the temp of the chip before it leaves the thermal transfer area, then your chip will get hotter.

in other words (again im a liquid cooling nub) you just need a powerful pump to make this work.
 
Well, the way I understand this, is you would probably see a little better temperatures depending on your radiator and flow rate. This would take the already rad-cooled water from your res, then put it equally to your two graphics cards. If anything, you would see better temps on the second graphics card since it didn't have to heat up going thru the first graphics card previously.

I am just starting into watercooling right now, but I don't really know how flow rates effect temps. I would think a higher flowrate would give lower temperatures since the water is going thru the waterblocks quicker to remove heat quicker. This would make your radiator have to work harder though by using higher fan speeds on it. I think your flow rate could be too high though since it would have to have enough time for the water to pass thru to remove enough heat if you used too big/fast of a pump. Is this correct?

That being said, splitting the two graphics cards up SHOULD give you one lower temperature water coming back into the loop after the 2nd Y-splitter if the two graphics cards are running at nearly the same temps. I don't know about temps cooling SLI cards or how the 1st cards temp compares to the 2nd, but I think the second card would heat up the water additionally because of the already hot water going onto the card's waterblock would steiffel it since you are trying to cool a card with warm/hot water.

If that is true, the radiator would have to work harder to dissipate the heat if it was in series vs. being in parallel. It seems you pretty much have to figure out the best flow rate for your setup to get the best cooling. You don't want the flow rate too high because your rad wouldn't dissipate the heat enough, and you don't want the flow rate too low because the waterblocks would soak the heat inefficiently.

^---- all of the above is me speculating.. I don't claim any of it to be true. Could anyone verify if some of it or all of it is correct? I am new to water cooling as well, building my first system as we speak. For now, I'm just cooling my CPU, but then when DX10 cards arrive, I'll go SLI and cool them. Parallel vs Series is the question. :confused:
 
Seems to me that the biggest objection to "parallel" flow loops has always been the difficulty providing equal flow to each component.
Normally, this would require a metering manifold to control waterflow and the extra flow reduction/complication this manifold introduces is simply not worth the effort.

In opty's case, switching to a series loop would remove a significant amount of tubing- the benefits of which would probably equal any gains a properly metered parallel loop would provide.

It's just a real world demonstration of the KISS principle.
 
sprocket said:
BTW, how difficult was it to fill the rez with it placed like that?
Click to expand...

I was thinking about trimming the tubing down a little in those areas, I was trying to eye ball how long the tubes needed to be while the computer was in 100 peices on my floor.

I filled and leak tested the system out of the case, so it was easy to fill the res, plus it just pulls out of that little bracket if I need to top it off.


MrWizard6600 said:
in other words (again im a liquid cooling nub) you just need a powerful pump to make this work.
Click to expand...

The pump I am using is the swiftech MCP655, rated at 317GPH, but definately not running at that now, but on high speed it should run at the max that loop is able to flow at.


w1retap said:
Well, the way I understand this, is you would probably see a little better temperatures depending on your radiator and flow rate. This would take the already rad-cooled water from your res, then put it equally to your two graphics cards. If anything, you would see better temps on the second graphics card since it didn't have to heat up going thru the first graphics card previously.
Click to expand...

The reason I split the flow was because the flow rate was extremely low, it was trying to push all the water through 1 little 1/4" tube and that was the bottleneck, by splitting it in 2 it effectively doubled the flow rate, because it seems that 2 1/4" tubes can flow as well as 1 3/8" tube. I was a little worried about not getting the same flow in each of the 2 1/4" blocks, but it seems that if the blocks are the same, and have the same flow restriction, that they will flow equally, give or take a small amount. But your theory about getting equal cooling to each graphics card is a pleasant side effect.
 
This will be a loop for the GPU's only
3/8's" tubing, Black Ice Pro II Radiator, 2x Panaflo 120 x 38_ 86.5CFM_ 35dba, MCP350 pump Modded... modified pump produces around 180. GPH flow rate (at 0.81 PSI head) while the original MCP350 only generated 110.7 GPH flow (at 0.36 PSI head). Now if the MCP350 can't produce adequete flow rates I do have a 2nd MCP655 that can used instead.

the tubing that goes from the Y 's to the GPU blocks will be equal in length.
I think once the tubing is full of water the flow rate to both would probably be close to equal
one of Danger Den's tech uses a Y style setup for his SLI rig. It's a bit different then what I'm looking at.
http://www.dangerdenstore.com/files/images/cases/wood/rwood/side off.jpg
 
You must log in or register to reply here.
Back
Top