Running 2 custom loops better?

[CrazyWavau]

Just wondering here, 

 

Most people run a single loop in their Water cooling system. Hypothetically, if your running OC everything and twin 10 series vid cards, the liquid temperature of the whole loop, is being affected by all your devices increasing the temperature across the board. But what if you had a single loop for the CPU and a separate loop for the video cards? would there be a discernable difference in performance?

[manikyath]

never having done any watercooling, i'd already see great benefit in splitting the loops just for the sake of bleeding air out the system.

[tp95112]

You would get less then a degree in change.Its more of asethetic thing only

[Shiv78]

3 minutes ago, manikyath said:

never having done any watercooling, i'd already see great benefit in splitting the loops just for the sake of bleeding air out the system.

this^ you could change CPU or GPU without having to drain the whole PC. However the more watercooling "stuff" you have, the greater chance of leaks.

[Not_Sean]

The theory of splitting them is solid but as mentioned above does very little,

 

The way guys deal with it is Big rad->GPU-Second Rad->CPU->pump->Res->back to big rad

 

Letting it cool in between does enough   

[manikyath]

Just now, Shiv78 said:

this^ you could change CPU or GPU without having to drain the whole PC. However the more watercooling "stuff" you have, the greater chance of leaks.

or.. if he'd put cpu block and both gpu blocks in series together with a bunch of rads, the increased pressure in the first set of seams light increase the chance of leaks.

 

not that i'm a qualified engineer tho...

[manikyath]

1 minute ago, Not_Sean said:

The theory of splitting them is solid but as mentioned above does very little,

 

The way guys deal with it is Big rad->GPU-Second Rad->CPU->pump->Res->back to big rad

 

Letting it cool in between does enough   

hasnt it already been proven by several people that the order doesnt really matter, because your PC quite honestly doesnt produce enough heat to make coolant temps sway so rapidly?

[stealth80]

1 minute ago, tp95112 said:

You would get less then a degree in change.Its more of asethetic thing only

wrong sorry, especially with SLI

 

I ran a 4790k and motherboard VRM's on one loop and 2x 970 GTX on another. The CPU loop was pretty cool most the time, the GPU dumped a SHIT TONNE OF HEAT out of the case - I literally had a 240mm radiator above my feet, and in the winter it was a foot warmer. Even though the CPU would sit around 44C in gaming and the GPU's around 53C, for some reason the GPU loop kicked a tonne of heat out, if that heat went into the CPU loop I could see the CPU been quite a bit more than 1C hotter.

 

 

[stealth80]

3 minutes ago, Not_Sean said:

The theory of splitting them is solid but as mentioned above does very little,

 

The way guys deal with it is Big rad->GPU-Second Rad->CPU->pump->Res->back to big rad

 

Letting it cool in between does enough   

doesn't make a difference in grand scheme, also pump before res is asking for BIG problems with air/bleeding and cavitating pump leading to early pump failure

[Enderman]

The total amount of rads of the whole system would be the same, so with two loops the only difference would be if you're running something that only uses the CPU or only uses the GPU.

In that case the GPU or CPU will run hotter but not affect the temperature of the other component.

Dual loops are done for more complexity for looks or for fun of building it, not for better temps.

If you care about temps then you should put all the components in the same loop.

[Shiv78]

3 minutes ago, stealth80 said:

wrong sorry, especially with SLI

 

I ran a 4790k and motherboard VRM's on one loop and 2x 970 GTX on another. The CPU loop was pretty cool most the time, the GPU dumped a SHIT TONNE OF HEAT out of the case - I literally had a 240mm radiator above my feet, and in the winter it was a foot warmer. Even though the CPU would sit around 44C in gaming and the GPU's around 53C, for some reason the GPU loop kicked a tonne of heat out, if that heat went into the CPU loop I could see the CPU been quite a bit more than 1C hotter.

Damn, nice rig!

[manikyath]

3 minutes ago, stealth80 said:

wrong sorry, especially with SLI

 

I ran a 4790k and motherboard VRM's on one loop and 2x 970 GTX on another. The CPU loop was pretty cool most the time, the GPU dumped a SHIT TONNE OF HEAT out of the case - I literally had a 240mm radiator above my feet, and in the winter it was a foot warmer. Even though the CPU would sit around 44C in gaming and the GPU's around 53C, for some reason the GPU loop kicked a tonne of heat out, if that heat went into the CPU loop I could see the CPU been quite a bit more than 1C hotter.

 

 

 

 

with SLI, going for a single loop also often looks messy. which if you're going custom watercooling isnt intended

[Enderman]

2 minutes ago, stealth80 said:

wrong sorry, especially with SLI

 

I ran a 4790k and motherboard VRM's on one loop and 2x 970 GTX on another. The CPU loop was pretty cool most the time, the GPU dumped a SHIT TONNE OF HEAT out of the case - I literally had a 240mm radiator above my feet, and in the winter it was a foot warmer. Even though the CPU would sit around 44C in gaming and the GPU's around 53C, for some reason the GPU loop kicked a tonne of heat out, if that heat went into the CPU loop I could see the CPU been quite a bit more than 1C hotter.

But then all the radiators in the CPU loop would also be cooling your GPUs.

So yes your CPU would be hotter but your GPUs would run cooler.

The overall temperature of the loop would balance out.

[Shiv78]

2 minutes ago, manikyath said:

with SLI, going for a single loop also often looks messy. which if you're going custom watercooling isnt intended

lol

[manikyath]

Just now, Shiv78 said:

lol

thats not SLI

[stealth80]

12 minutes ago, Enderman said:

But then all the radiators in the CPU loop would also be cooling your GPUs.

So yes your CPU would be hotter but your GPUs would run cooler.

The overall temperature of the loop would balance out.

no it wouldn't, I tried it, hence why i'm commenting on this topic. Theres only so much you can achieve with regards to cooling vs ambient temps, theres a point where the gains of extra rads/surface area have little to no benefit due to the difference between loop/coolant temp vs ambient temperatures. I had a 360mm x 60mm radiator for CPU and VRM's and a 360 x60mm + 240 x 60mm for 2x 970's. In this case it was better for me to not have the extra heat from the GPU's going to the CPU loop, CPU temps were cooler, GPU temps remained the same - keep in mine 360mm + 240mm is more than most people use for a full loop, not an isolated GPU loop. 

 

Also keep in mind that by having dual loops you are able to control fan speeds for CPU and GPU temps individually

 

12 minutes ago, Shiv78 said:

Damn, nice rig!

Thanks, It was, it was too big though and I had to downsize, current rig:

 

[Shiv78]

6 minutes ago, manikyath said:

thats not SLI

true

[Shiv78]

Just now, stealth80 said:

no it wouldn't, I tried it, hence why i'm commenting on this topic. Theres only so much you can achieve with regards to cooling vs ambient temps, theres a point where the gains of extra rads/surface area have little to no benefit due to the difference between loop/coolant temp vs ambient temperatures. I had a 360mm x 60mm radiator for CPU and VRM's and a 360 x60mm + 240 x 60mm for 2x 970's. In this case it was better for me to not have the extra heat from the GPU's going to the CPU loop, CPU temps were cooler, GPU temps remained the same - keep in mine 360mm + 240mm is more than most people use for a full loop, not an isolated GPU loop. 

 

Thanks, It was, it was too big though and I had to downsize, current rig:

 

 

yeah, saw in your sig. Still sexy and now more unique color scheme

[Enderman]

27 minutes ago, stealth80 said:

In this case it was better for me to not have the extra heat from the GPU's going to the CPU loop, CPU temps were cooler, GPU temps remained the same

This goes against all physics and thermodynamics laws that exist.

[Mick Naughty]

If you have the room go for it. If anyone has had a single loop and only ran the gpu's you would realize how much hotter they can be over a single cpu. Played a game last night that I don't normally play, forgot my cpu can sit at idle temps while my cards bake because the fans never increased their speed as they are based off cpu temp. Gonna put another pump in my rig if I continue to play games.

[stealth80]

1 hour ago, Enderman said:

This goes against all physics and thermodynamics laws that exist.

no it doesn't, ive studied thermodynamics as part of my aeronautical engineering degree, the transfer can only exist between two fluids and that transfer depends on the co-efficient of transfer between materials (radiator) and the temperature difference between the fluid in the loop and the ambient. it's pretty much impossible to get the fluid temperature to ambient levels once heated without completely removing the heat source, as I said, even with a 360 and a 240mm the heat was still more than that of the CPU loop with a single 360mm, so there was no point in dumping the GPU heat into the CPU loop just to increase the temp of the CPU for no drop in GPU temp - the only way to make it more efficient at this point was to reduce the ambient which isn't possible without a weather machine 

[stealth80]

1 hour ago, Mick Naughty said:

If you have the room go for it. If anyone has had a single loop and only ran the gpu's you would realize how much hotter they can be over a single cpu. Played a game last night that I don't normally play, forgot my cpu can sit at idle temps while my cards bake because the fans never increased their speed as they are based off cpu temp. Gonna put another pump in my rig if I continue to play games.

This is my point and another that agrees

1 hour ago, Enderman said:

This goes against all physics and thermodynamics laws that exist.

 

[Enderman]

10 minutes ago, stealth80 said:

no it doesn't, ive studied thermodynamics as part of my aeronautical engineering degree, the transfer can only exist between two fluids and that transfer depends on the co-efficient of transfer between materials (radiator) and the temperature difference between the fluid in the loop and the ambient. it's pretty much impossible to get the fluid temperature to ambient levels once heated without completely removing the heat source, as I said, even with a 360 and a 240mm the heat was still more than that of the CPU loop with a single 360mm, so there was no point in dumping the GPU heat into the CPU loop just to increase the temp of the CPU for no drop in GPU temp - the only way to make it more efficient at this point was to reduce the ambient which isn't possible without a weather machine 

Who said anything about ambient temperatures???

You're just making stuff up at this point.

 

If you have two loops, and the GPU loop temp is higher than the CPU loop temp, if you combine both together the GPU temps will go down and CPU temps will go up.

This is literally grade 1 math.

 

9 minutes ago, stealth80 said:

This is my point and another that agrees

That's completely different, that's not what we're talking about here...

We're not talking about fan speeds or anything like that.

Yes GPUs will run hotter than a CPU while playing most games, that's common sense.

[stealth80]

11 minutes ago, Enderman said:

Who said anything about ambient temperatures???

You're just making stuff up at this point.

 

If you have two loops, and the GPU loop temp is higher than the CPU loop temp, if you combine both together the GPU temps will go down and CPU temps will go up.

This is literally grade 1 math.

 

That's completely different, that's not what we're talking about here...

We're not talking about fan speeds or anything like that.

Yes GPUs will run hotter than a CPU while playing most games, that's common sense.

Of course ambient matters! A loop can't cool below ambient fact. And again, look it up, theres a point where it doesn't make a difference how much surface area there is, the heat cant be removed due to the difference between ambient and fluid/loop been too small. Its fact and you don't do thermo's in grade one math, don't speak trash. The transfer is based on delta between the loop and ambient, the co-efficient of the materials used to transfer the heat : so in this case, the radiator fins, the air (from the fans) and the coolant used. This is basic thermos, then you need to consider the amount of airflow and the temperature (delta) of the airflow from the fans used. 

 

Also, like I said, I tried it, the temps didn't change, the GPU's were still 53C! Yes the coolant temp would increase in a single loop (in my case very slightly 360x2 + 240), but it was still cooler than that of the CPU so the CPU temps didn't increase

 

[Enderman]

7 minutes ago, stealth80 said:

Of course ambient matters! A loop can't cool below ambient fact. And again, look it up, theres a point where it doesn't make a difference how much surface area there is, the heat cant be removed due to the difference between ambient and fluid/loop been too small. Its fact and you don't do thermo's in grade one math, don't speak trash. The transfer is based on delta between the loop and ambient, the co-efficient of the materials used to transfer the heat : so in this case, the radiator fins, the air (from the fans) and the coolant used. This is basic thermos, then you need to consider the amount of airflow and the temperature (delta) of the airflow from the fans used. 

 

Also, like I said, I tried it, the temps didn't change, the GPU's were still 53C! 

 

First of all, ambient temperature would affect both loops equally, so it literally does not matter for this equation.

It is a fact that if one system has more heat energy than another, and you join the two, the resulting heat energy density of a system will be between the two original values.

 

Maybe you had too many radiators to make a difference, or maybe your ambient changed when you measured the second time, since you probably measured them at very different times because rebuilding an entire loop isn't super fast.

 

Maybe you didn't run a constant workload, idk, you don't seem to have run any proper tests.

 

The point is, a single loop will share heat evenly across all radiators, which will give better temps (for the component being stressed) when only stressing CPU or GPU.

And unless you have an unbalanced ratio of radiators to component heat output, it will give the exact same temps as a dual loop system when stressing all components at the same time.