My first watercooled build: 9980XE, RTX 2080Ti, etc.

VanGoghComplex said:
I've never tried to clean tubing, and maybe I'm way off base here (my chemistry knowledge is minimal), but I'm betting that boiling that soft plastic in a mild acid is probably what clouded it.

I personally have used Primochill LRT on my soft tubing builds and never had problems, but I've heard good things about Mayhems too.
Click to expand...

Awesome, I'll just replace it with new Mayhems. Luckily I got a big spool, so I should be set.
 
I don't clean tubing. It gets cloudy eventually. All of it does. I just replace it every so often and move on with my life. Your other choice is zero maintenance tubing like what I have in my rig.
 
jyi786 said:
Update: tried cleaning out the tubes last night. They cleaned out well with boiling water and vinegar, but clouded up. Not sure I like them anymore, so I'll just be replacing them.

I've read that the Mayhems Ultra Clear tubing, which I'm using, doesn't have plasticizer in it, so I think I should be good? Just trying to prevent having to do this all over again.
Click to expand...

Don't use clear tubing... you only invite bio growth with clear anyways.
 
thesmokingman said:
Don't use clear tubing... you only invite bio growth with clear anyways.
Click to expand...

This is because sunlight, right?

I guess for inside the case is ok because there is no light, but outside the case I might have to go with ZMT.
 
Cleaning completed. Yesterday I tore down all the waterblocks and cleaned them thoroughly. The CPU block had something strange in it. It looked like black feathery fuzz. Not sure where that came from, but it washed right out with a toothbrush. The GPU waterblock was a pain in the rear to remove. MSI puts these semi-permanent stickers over the enter waterblock that are a PITA to remove. Once I got them off, it was smooth sailing. Cleaned everything, put it back together. Leak tested, ran and flushed with 2.5-3 gallons of distilled water, then put in 12 drops of the biocide.

In service and running very well now. The D5 pump noise signature is drastically altered; it is much quieter because of the way I configured their locations. I am keeping them on speed setting 5; oddly lowering them to 3 is louder to my ears.

The next steps for me are to upgrade to ZMT tubing externally, get one or two more elbows, and at least another pair of Koolance QDCs for the tubing so that it is fully modular.
 
I did. It made very good contact with the waterblock, and I checked with a flat razor blade. No concave/convexness to report, at least not that I can see.
 
German Muscle said:
did you drop any temp off the flush?
Click to expand...

You mean did temps drop? Yes, they did. Specifically, seems the CPU has dropped temps. I can Aida64 FPU test the thing and it'll hold at around 90C, whereas before it would hold at 101C. Additionally, it stays about 10C cooler at idle, pretty much the same as the water temp and GPU temp. Prior, there used to be a ~10 degree delta in there.
 
What's there head pressure when using dual d5's? How far can you place the rad from the tower before you run into issues?
 
Blue4130 said:
What's there head pressure when using dual d5's? How far can you place the rad from the tower before you run into issues?
Click to expand...

No idea. Right now, the rad is about 10 feet away from my computer. I've heard of people putting their rad on a different floor with only ONE D5. With duals, I don't know, but conventional wisdom tells me that I'll be able to place it far. Really far.
 
jyi786 said:
No idea. Right now, the rad is about 10 feet away from my computer. I've heard of people putting their rad on a different floor with only ONE D5. With duals, I don't know, but conventional wisdom tells me that I'll be able to place it far. Really far.
Click to expand...

Yeah, I have a D5 pushing from my basement (6’ down) to my 9900k/2080ti (~600watts) through 1/4” ID tubing no problem. I used a 12 gallon storage tub as a resevoir.

Once the loop is filled head loss is effectively zero, but that little pump has to problem with the initial push.
 
So over the past few days I had been struggling with high temps on both the GPU and CPU. GPU would hit around 70C in gaming, and CPU would hit 80C.

I discovered that I had to clean the GPU block out again after I noticed some debris in the cooling fins on the GPU surface area (don't know how the hell it got in there). I also found that the GPU block wasn't mounted correctly (my fault: the screws needed to be torqued down much tighter). Then the CPU was getting ridiculously high temps and the computer would shut down, so I had to clean the block, make sure I mounted the waterblock base correctly. I ran out of thermal paste, so I had to wait for that to be delivered. When it was, while I was at it, I decided to convert from parallel to serial flow.

What did this net me? The following:

CPU/GPU @ idle: idle at water temp level, which is close to ambient by 1-2 degrees
CPU @ Aida64 FPU load + GPU Furmark (running simultaneously): CPU 101C and GPU 50C
CPU @ regular CPU fully loaded + GPU Furmark (running simultaneously): CPU 70C and GPU 48C

In real life workloads, like regular gaming (Destiny 2, Metro Exodus, etc.), CPU temp never exceeds 55C and GPU never exceeds 48C. Metro Exodus is a perfect test for this, as I've found that game uses ALL available threads (I have 36, and there are times they are pegged at 100%). My rig is completely silent while gaming with all details maxed.

My advice for everyone is to only do parallel flow if you absolutely have to, and do your very best to adhere to the inlet flow pattern for the GPU. I read tons of articles where it states that there is a minuscule to no difference in temps between parallel and serial, or using the "outlet" as the inlet on the GPU, so I thought switching from parallel to serial would do nothing for me. In my case, switching the GPU to serial and thereby using the left port as the inlet and right as outlet netted me a 4C improvement. That's a lot for a system like mine. Additionally, converting the CPU from parallel to serial flow netted me at least 5C shaved off the top. In my case, those 5 degrees are the difference between stability and the system just completely shutting down.

I've pretty much achieved what I set out to do with this build, and I'm finally happy. Now to enjoy it for the next 10 years or so. :)
 
mac1 said:
Looks like mad scientist project. Cool, no pawn intended...
Click to expand...

Thanks! And so far, it's performed like a mad scientist project. Keeps the beast ultra cool. In between then and now, I've gotten the VRMs under water as well. Idle for all components is ambient room temp. Full load doesn't exceed 50C on all components, ever, unless I do a load like OCCT or Prime95.
 
You must log in or register to reply here.
Back
Top