Y-split to cpu and gpu vs serial connection

[FUSER]

I have 3 radiators, one gpu , one cpu, planning to overclock.

i am looking at the following loop design:

 

reservoir -> 360mm radiator #1 -> 360mm radiator #2 -> ysplit:

- Split #1 to cpu

- split #2 to 420mm radiator -> GPU
 

GPU out and CPU out combine with reverse ysplit to reservoir inlet.

 

i know it is more work than a serial loop and potentially the cooling gains are minimal if anything. Will this cause under flow to my GPU since the water has to pass through the 3rd radiator as well?

shall i just Y-split after the 3rd radiator to cpu and gpu or just use the gpu block as a parallel loop with the cpu?

 

thank you

[LIGISTX]

6 minutes ago, FUSER said:

I have 3 radiators, one gpu , one cpu, planning to overclock.

i am looking at the following loop design:

 

reservoir -> 360mm radiator #1 -> 360mm radiator #2 -> ysplit:

- Split #1 to cpu

- split #2 to 420mm radiator -> GPU
 

GPU out and CPU out combine with reverse ysplit to reservoir inlet.

 

i know it is more work than a serial loop and potentially the cooling gains are minimal if anything. Will this cause under flow to my GPU since the water has to pass through the 3rd radiator as well?

shall i just Y-split after the 3rd radiator to cpu and gpu or just use the gpu block as a parallel loop with the cpu?

 

thank you

Don’t split it. With that much rad space, there is effectively 0 reason to worry about this. 
 

I run a 45mm 420 and a 60mm 280, and my GPU runs at ~51c in games, and CPU is in the low 60’s….. with my fans at ~800-900 rpm.

 

The main point of custom water is to be quiet, and with that much rad space, you will be. Trying to save 1 c here or there just doesn’t matter and is a fools errand. Don’t add complexity to an already complex system. 

[fonzz1e]

afaik 2 split flow will result in 1 component getting lower flow rate

• split#1 has less flow restriction, split#2 will get less flow
• same vice versa when split#2 has more restriction than split#1

parallel flow setup is similar, but it's done for easier tubing runs

 

but overall its up to u, if yer using soft tubing yer can always try diff flow setups & compare its temp result

[FUSER]

thank you for the insight.

 

[FUSER]

21 hours ago, LIGISTX said:

Don’t split it. With that much rad space, there is effectively 0 reason to worry about this. 
 

I run a 45mm 420 and a 60mm 280, and my GPU runs at ~51c in games, and CPU is in the low 60’s….. with my fans at ~800-900 rpm.

 

The main point of custom water is to be quiet, and with that much rad space, you will be. Trying to save 1 c here or there just doesn’t matter and is a fools errand. Don’t add complexity to an already complex system. 

is there any benefit doing the split after radiator #3 and send simultaneously half flow to CPU and half flow to GPU before converging them back into one tube in the reservoir inlet (see picture, TR=top radiator, SR=side radiator, FR=front radiator).

 

[LIGISTX]

24 minutes ago, FUSER said:

is there any benefit doing the split after radiator #3 and send simultaneously half flow to CPU and half flow to GPU before converging them back into one tube in the reservoir inlet (see picture, TR=top radiator, SR=side radiator, FR=front radiator).

 

Realistically, what benefit are you looking to get? The GPU is already going to be in the 40’s, maybe 50’s depending on ambient. And the CPU is more or less limited by IHS thermal transfer efficiency.

 

The amount of rad space you will have is way overkill (which is totally fine), and without trying to do any weird multi path ideas will be more then plenty to be silent under load. A few degrees here or there is within measurement error for most setups, and realistically doesn’t matter at all.

 

Make the loop 1 of two things:

1) Easy to maintain/easy to setup

2) Esthetically pleasing to you

 

Optimal flow path is really not a concern. The temperature of the water in the loop is pretty similar across the entire flow path, so order of components doesn’t matter much. The only thing that does matter somewhat is if you have a radiator acting as intake, put that after the radiator/s acting as exhaust. This helps reduce pre-heating the air a little bit and thus not reducing the efficiency of the water to air heat exchange happening in the radiator. 

[FUSER]

9 hours ago, LIGISTX said:

The only thing that does matter somewhat is if you have a radiator acting as intake, put that after the radiator/s acting as exhaust. This helps reduce pre-heating the air a little bit and thus not reducing the efficiency of the water to air heat exchange happening in the radiator. 

Right now all my radiators i am planning to put as both intake/exhaust (depending on how you read the loop), meaning they are all either behind cpu/gpu or after them. Are you saying keep one rad before cpu/gpu and 2 rads right after ?

[LIGISTX]

6 hours ago, FUSER said:

Right now all my radiators i am planning to put as both intake/exhaust (depending on how you read the loop), meaning they are all either behind cpu/gpu or after them. Are you saying keep one rad before cpu/gpu and 2 rads right after ?

No, intake means it’s sucking in cool air from outside the case and blowing warm air into the case, exhaust means it’s blowing it’s hot air out of the case.

 

Radiators work by exchanging heat from the water flowing through them with the air being blown over the fins. The air that exits the radiator is now warm. So if you put the first radiator after a heat producing component as intake, your dumping more heat into the case then if you make it the last radiator in the loop.

 

 

[FUSER]

2 minutes ago, LIGISTX said:

No, intake means it’s sucking in cool air from outside the case and blowing warm air into the case, exhaust means it’s blowing it’s hot air out of the case.

 

Radiators work by exchanging heat from the water flowing through them with the air being blown over the fins. The air that exits the radiator is now warm. So if you put the first radiator after a heat producing component as intake, your dumping more heat into the case then if you make it the last radiator in the loop.

 

 

Gotcha. My front radiator is away from all components and is bringing hot air inside, but is right next to the side radiator that pushes air from inside to outside to compensate. Whatever remains warm will then go to the 3rd radiator (top) which also pushes air from inside to outside again. 

will this be good enough?

 

I will connect the cpu and gpu after the 3rd radiator (still wondering whether to ysplit to gpu+cpu for aesthetics mostly or go cpu->gpu either serially or parallely)

 

[LIGISTX]

41 minutes ago, FUSER said:

Gotcha. My front radiator is away from all components and is bringing hot air inside, but is right next to the side radiator that pushes air from inside to outside to compensate. Whatever remains warm will then go to the 3rd radiator (top) which also pushes air from inside to outside again. 

will this be good enough?

 

I will connect the cpu and gpu after the 3rd radiator (still wondering whether to ysplit to gpu+cpu for aesthetics mostly or go cpu->gpu either serially or parallely)

 

What I am saying is the front radiator which is an intake, should be the one “farthest away” from the heat generating parts of the loop. That way the liquid passing through it has already been cooled by 2 other rads. So it should be “heat sources”, rad 1 (top or side), rad 2 (top or side), rad 3 which is the front. 

[FUSER]

28 minutes ago, LIGISTX said:

What I am saying is the front radiator which is an intake, should be the one “farthest away” from the heat generating parts of the loop. That way the liquid passing through it has already been cooled by 2 other rads. So it should be “heat sources”, rad 1 (top or side), rad 2 (top or side), rad 3 which is the front. 

Not sure if that makes sense since my front rad takes cool air from outside to cool the rad which is receiving the theoretically “warmer” flow…

Current orientation:

front rad cools by sucking from outside in

side rad cools by blowing from inside out

top rad cools by blowing from insude out

[LIGISTX]

30 minutes ago, FUSER said:

Not sure if that makes sense since my front rad takes cool air from outside to cool the rad which is receiving the theoretically “warmer” flow…

Current orientation:

front rad cools by sucking from outside in

side rad cools by blowing from inside out

top rad cools by blowing from insude out

I’d watch the video I posted… 

 

Technically the most efficient thermal transfer would happen if you pump the warmer water through the rad seeing the coolest air. But in a multi rad system, if your blowing warmer air into the case and this over the other radiators, both other rads will work less efficiently and would hurt the loops performance. Again, watch the video… 

 

Make the rad acting as intake the last rad in the loop. And again, I think you are way over thinking this. The amount of rads your planning to use will easily cool your system with the fans spinning near inaudibly. I’m just trying to help provide the best layout that lets all 3 rads, as a system, work at their best. Thermodynamics dictates the front rad being the first in the loop would increase the efficiency of that rad, but it will reduce the efficiency of the other two… which as as the system as a whole is less efficient. 

[FUSER]

On 12/1/2022 at 12:22 PM, LIGISTX said:

I’d watch the video I posted… 

 

Technically the most efficient thermal transfer would happen if you pump the warmer water through the rad seeing the coolest air. But in a multi rad system, if your blowing warmer air into the case and this over the other radiators, both other rads will work less efficiently and would hurt the loops performance. Again, watch the video… 

 

Make the rad acting as intake the last rad in the loop. And again, I think you are way over thinking this. The amount of rads your planning to use will easily cool your system with the fans spinning near inaudibly. I’m just trying to help provide the best layout that lets all 3 rads, as a system, work at their best. Thermodynamics dictates the front rad being the first in the loop would increase the efficiency of that rad, but it will reduce the efficiency of the other two… which as as the system as a whole is less efficient. 

Thank I will route the rads as you suggested.

my other question was: the output from rad#3 will it be an issues flow-wise y-splitting it in an effort to reach cpu and gpu simultaneously (instead of connecting them serially) ?

same goes with cpu and gpu outputs combining into 1 tube  before entering the front rad. Will these be a problem flow-wise?

[LIGISTX]

1 hour ago, FUSER said:

Thank I will route the rads as you suggested.

my other question was: the output from rad#3 will it be an issues flow-wise y-splitting it in an effort to reach cpu and gpu simultaneously (instead of connecting them serially) ?

same goes with cpu and gpu outputs combining into 1 tube  before entering the front rad. Will these be a problem flow-wise?

Again, I really advise against parallel flow. It just doesn’t matter. What is your goal in using parallel flow? I assume the answer would be “so the parts run cooler”. The difference might be 1-2c at best, while it adds complexity and points of failure for almost no gain.