Is this wrong?

[WildArmsfx]

I was surfing a new website for builds when I saw this. Looks like a 3 way split was used to supply both the cpu and gpu intake/outake. I imagine this will impeded flow to either one that has the most resistance, like bypassing the cpu and just the majority of the water going thru gpu.

Im not trying to mock this person, just seemed off to me. If Im going to warn him something is wrong I want to be educated as to why.

 

[godevskii]

It's not wrong, there's some downsides of doing it like this, one is it will take longer for the air to leave the loop after initial fill and it will exert more pressure on the exit points making the pump having to work harder, but since this is a relatively small loop i see no problems.

 

[Mr.Scott]

Definitely not the preferred method. OP is correct, water will follow the path of least resistance.
One of the two components will be getting short changed on the cooling unless the block restriction is similar.

 

[Zerileous]

How is this different from the many parallel CPU / GPU loops that enter the GPU from the bottom on the entry side, then exit from the top on the entry side to the CPU block's intake, then exit the CPU down to the top exit side of the GPU, and finally exit the GPU from the bottom GPU exit side? Sorry if that is confusing. Random example from GIS:
.

Maybe I don't understand how the part that has the ports on a GPU block works. I noticed there is a baffle inside mine from the top to bottom ports, but I didn't disassemble it. I assumed the block it self was not split in any way, so that some of the flow that was directed into the block then was spit back into the top port in these setups. Is the baffle that important?

I also understand how SLI of multiple identicle blocks in paralell would get equal flow, but if CPU blocks are typically significantly more restrictive, wouldn't the coolant follow the path(s) of least resistance and decrease flow to the CPU? Essentially I would see it as shunting fluid away from the CPU block via the less restrictive GPU blocks. Always wondered about this, but it didn't come up in my loop design. Would this be any different in the image posted above vs the one in the OP?

 

[godevskii]

It's not different, but it has been proven to work. Still has @Mr.Scott says "Definitely not the preferred method" and it might perform a bit worse and with some downsides. I'm no scientist but i think as soon as the liquid reaches some part and fills it, pressure starts to build and then it's got to come out throw the less restrictive way, so that also helps the flow going.

 

[Mr.Scott]

I agree. Probably fine. Only a couple degree worse temps is possible. Just not the preferred method is all.
Aesthetically, looks great.

 

[Zerileous]

I wonder if anyone has measured the flow exiting a restrictive CPU block vs the flow exiting the GPU block(s) in a setup like this. I'm sure the water flows through the more restrictive block, but I'd assume at a lower rate. Maybe the more restrictive components are less dependent on flow rate? I've heard 1gph or 1.5gph as a general guideline, and I know it has been tested, but I'm not familiar with the actual test results/differences.

 

[WildArmsfx]

I would definetly not do this BTW. Its a recipe for disaster. I was thinking he may have been using a special splitter to make it work, he did not and that type of splitter does not exist for water cooling.

I think we all agree here but are having different bars for what equates to "wrong". To me less efficient with the increased chance of error is suitable to mark something as wrong. But to others wrong seems to mean "doesn't end in catastrophic failure". Which both are fair.

If you like a design/look over efficiency and reliability, then ok.

 

[1pqlamz]

I had a similar setup which had 2 980s in sli, with fluid entering and exiting from bottom. I had the pumps pwm setup according to CPU temp, the result is, the flow to the top GPU becomes so little that causes the GPU to overheat and ultimately deformed the GPU block due to the overheating. And there goes my GPU block, lucky the leak didn't cause any further damage, so in short DO NOT setup parallel loop unless you are sure that all blocks are going receive sufficient flow all the time.