Pump and tubing for GTX 980 Ti loop

D

Deimos

[H]ard|Gawd
Joined
Aug 10, 2004
Messages
1,166
In the wake of the lackluster Fury X I'm dropping some coin on a 980 Ti SLI setup with a second custom loop.

Water Cooling availability is fairly sparse at the arse end of the world so I pretty much have no choice but to go for a full EKWB setup (although I'm not unhappy about that) I'm just stuck in terms of the type of pump and tubing to get.

Am I correct in assuming the best option pump wise is a DDC type pump (for higher static pressure vs high flow D5)? I'm looking at getting an EK-SBay Dual DDC for this loop.

Tubing wise, I have 1/2" ID tubing for the CPU but was looking at going to 3/8" ID for the second loop. I'm assuming flow will be highly restricted by the VGA blocks so the tubing diameter is less important.

Any thoughts would be greatly appreciated.
 
All i can say is I've been using 1/2" ID x 5/8" OD tubing with SLI setups since 2009 and it hasn't failed me yet.though I'm in series with the CPU.

running dual 980s now.
 
Yeah I have 1/2" ID on my CPU but the block and rad are not particularly flow restrictive and I'm running a high flow D5 pump. I'm not changing this setup any time soon, although I'm probably overdue for a new CPU block.

I read a recent post on here where a forum member took time to test a number of different diameter tubing and found no difference in temps and flow rates, he also tested a dozen or so different flow rates by changing pump speed and found only a 1-2C delta from lowest to highest.

As far as I know, a full VGA block is high flow resistance and with two in a loop it would be better to go for DDC type pump for high static pressure, so this is my primary concern. If going down an ID size won't make any significant difference to flow rate I would prefer to go smaller for ease of installation.
 
i run a crossfire setup with the GPU's and CPU all in parallel on 3/8" ID and it has not been an issue. temps remained the same when the MCP35x was switched for a D5-PWM (though the D5 is silent in comparison).

personally i think even the 3/8" ID is overkill, just finding 1/4" that won't kink is difficult and costly enough not to make it worth pursuing. i swear a lot of automotive auxiliary coolant pumps (for use in hybrids/EV and air/water intercooling systems) are d5 based, and they are tasked with moving considerably larger volumes, under much greater thermal loads.
 
I would stick with the D5. The D5 is generally more reliable than the DDC.

It has been proven there is minimal flow rate difference in a typical watercooling setup between 3/8" ID and 1/2" ID. There is some flow rate penalty with 1/4" ID.
 
Tsumi said:
I would stick with the D5. The D5 is generally more reliable than the DDC.

It has been proven there is minimal flow rate difference in a typical watercooling setup between 3/8" ID and 1/2" ID. There is some flow rate penalty with 1/4" ID.
Click to expand...

I ended up going for the D5, the sales person said the D5 is more reliable, I can't disagree since my current D5 has lasted me 8 years so far! There should be no problem with a restrictive loop, will running the cards parallel help in this regard?
 
When you go parallel, unless you have everything set up just exactly so, you can't easily control how much flow goes through the cards -- one card may get a lot, the other not so much, and you'd see that as a temperature difference between the cards depending on how bad the flow imbalance is.

On the other hand, running them in series, the second card will always be a tiny bit hotter than the first card, but it will be by a predictable and repeatable amount, and not terribly high as long as you have sufficient flow (and with 1/2" tubing and that pump, you would).
 
Deimos said:
I ended up going for the D5, the sales person said the D5 is more reliable, I can't disagree since my current D5 has lasted me 8 years so far! There should be no problem with a restrictive loop, will running the cards parallel help in this regard?
Click to expand...

It would. Then again, you have enough pumping power that it doesn't really matter.

Brian_B said:
When you go parallel, unless you have everything set up just exactly so, you can't easily control how much flow goes through the cards -- one card may get a lot, the other not so much, and you'd see that as a temperature difference between the cards depending on how bad the flow imbalance is.

On the other hand, running them in series, the second card will always be a tiny bit hotter than the first card, but it will be by a predictable and repeatable amount, and not terribly high as long as you have sufficient flow (and with 1/2" tubing and that pump, you would).
Click to expand...

Identical blocks have identical restriction, and will have identical flow in a typical parallel GPU setup.
 
Yep, that's what I see in actual usage, and mine are even in parallel with the CPU. The GPUs run about 10c hotter but the same between them. The other advantage to fully parallel is that each new block reduces overall restriction. A hose clamp can be used to balance the flow rate, though I have not needed it
 
Got it all up and running.
Surprisingly I can still hit 80C if I run Valley at 4k while overclocked and running the fans at 600rpm, bump the fans to 1000rpm and the temps drop quite dramatically to around 60C

There is only a 3-4C delta between the two cards which may be due to differences in the cards rather than the cooling.

All in all I'm really happy with it, will post pics when I have time to take them.
 
Tsumi said:
What radiator are you using?
Click to expand...

It's a EKWB EK-CoolStream PE 240 Radiator

Killerxp100 said:
That is quite the difference for a 400rpm jump.
Click to expand...

It's 5 high static pressure fans (3 on the CPU rad, 2 on the VGA rad) in a push pull config. CPU is around 35-40C while gaming so increasing fan speed makes the biggest difference to the video cards. Also lowering the fan speed means more of the CPU heat ends up in the video card rad because it has more time to heat up the air before it gets there.
 
I've always used 5/8" tubing.

Flow rate between the different sizes of tubes isn't really a problem in most setups.
 
You must log in or register to reply here.
Back
Top