Single GPU and Single CPU in a Parallel Flow Loop?

Avathar77 · 16 Jun 2012 at 23:08

I have only seen parallel flow loops in dual GPU configs, but can this be done in a single CPU and GPU setup?

This is the idea

According to what I have read, some say the restriction in GPU bock is lower than CPU block and this will cause majority of the coolant to go through the GPU block while starving the CPU block!

Also, serial is the easiest way to go, I know that, but I think the above way will give the neatest tube routing.

Any thought on this?

malfunkshun · 16 Jun 2012 at 23:11

As you have read, water will take the path of least resistance so, unless the two blocks are equally restrictive, you will find one gets more coolant than the other.
IMO you also need the tubing to be as short as possible, if you can balance that with it being very tidy then that would be a bonus.

Avathar77 · 16 Jun 2012 at 23:17

Do you mean the tubing between GPU - out and CPU-in?
I know my CPU block is much more restrictive than the GPU block(from reviews)

malfunkshun · 16 Jun 2012 at 23:21

Avathar77 said:
Do you mean the tubing between GPU - out and CPU-in?

I mean the tubing in general, less is better.

Avathar77 said:
I know my CPU block is much more restrictive than the GPU block(from reviews)

Indeed, in which case you will have flow balancing issues between the parallel sections of the loop.

Avathar77 · 16 Jun 2012 at 23:24

[URL="

This is how my loop is right now, so tubing is minimal

I don't like the tube from CPU-out to RES-in, that is why I wanted to go for an alternative loop.
Back to drawing board then

rjk · 17 Jun 2012 at 00:30

It will work like that. I had mine running like that for a while and plan on doing it in my next build

Avathar77 · 17 Jun 2012 at 00:35

rjkoneill said:
It will work like that. I had mine running like that for a while and plan on doing it in my next build

I was hoping you would see this thread

I know you have done your TJ07 build with a parallel loop, but you had 2 GPUs in it with the CPU! So for you it is justifiable, but I am thinking if I am tring to make things too complected or not

How was the temps when you went back?

KiiYzOo · 17 Jun 2012 at 00:36

Wouldn't this be exactly like the big res theory? although, imo, this would be much more practical, as it makes it look cleaner

and by the big res theory, I mean that after a while, flow rates would even out?

Avathar77 · 17 Jun 2012 at 00:41

Isn't the big res theory about temps evening out?

KiiYzOo · 17 Jun 2012 at 00:45

I meant it as in the flow rates are like the temps in that theory

Cold_And_Miserable · 17 Jun 2012 at 06:36

Avathar77 said:
This is the idea

Any thought on this?

That diagram shows a series circuit or at least it'll just flow through the GPU block which is normally much less restrictive

This is parallel:

Uploaded with ImageShack.us

Avathar77 · 17 Jun 2012 at 06:39

It works out to be the same

Why do you have 2 pumps? for redundancy?

Cold_And_Miserable · 17 Jun 2012 at 06:47

Avathar77 said:
It works out to be the same
Why do you have 2 pumps? for redundancy?

2 pumps in series should increase the flow rate and provide redundancy.

A parallel loop with 1 pump per block drawing water from the same reservoir might work but theres no theoretical or practical advantage.

GoogalyMoogaly · 17 Jun 2012 at 17:06

PanMaster said:
That diagram shows a series circuit or at least it'll just flow through the GPU block which is normally much less restrictive

This is parallel:

Uploaded with ImageShack.us

It's not really series as it doesn't do one and then the other, it does one or the other. It's much like what you drew except the split is at the graphics card, not before it.

Jokester · 17 Jun 2012 at 17:13

CPU blocks tend to like high flow, so putting them in parallel with a low restriction GPU block which will generally take more of the flow will give you worse temps. How much worse? Hard to say put it won't be that much.

It would be interesting to see a restriction comparison of a typical full cover GPU block against CPU blocks. If the GPU block is substantially less restrictive if might give a fairly large increase in CPU temperatures (5-10degC) if the flow through the CPU drops substantially below that critical value of about 1gpm. All gues work though, if the CPU flow is still high enough the difference could only be about 1degC, if that.