Custom Water Loop Performance Increases

[JT241]

Hey everyone, I'm considering doing my first custom water loop I was wondering if anyone had evidence of whether or not a custom water loop actually increases CPU performance over either AIO or heavy air cooling. I know LTT came out with a video not too long ago comparing AIO's vs air cooling with air cooling coming out on top, but how does that compare to custom water loops?
Also in a much older video Linus did mention that you shouldn't expect much of a CPU performance with a custom loop as its desired more for aesthetics and noise reasons, but I found that somewhat hard to believe. Any thoughts?

[KarathKasun]

Unless you are pushing system killing voltages, the performance boost is minimal.

 

Without overclocking, the benefit is limited to slightly higher GPU boost speeds.

 

With extreme air you can hit similar OC speeds as well.

[Skiiwee29]

Custom water loops is almost all in it for the looks and aesthetics, not the actual performance. You do gain minimal over top end air coolers or high end AIOs, but truthfully, not enough to really justify the cost compared. 

[KarathKasun]

Just now, Skiiwee29 said:

Custom water loops is almost all in it for the looks and aesthetics, not the actual performance. You do gain minimal over top end air coolers or high end AIOs, but truthfully, not enough to really justify the cost compared. 

Yep, worst price/perf PC upgrade.

[Enderman]

A loop only changes CPU temperatures, not performance.

The performance is affected by whether you overclock or not, assuming you have no throttling in the first place.

 

If the custom loop has more radiator area than the AIO you're comparing it to, then it has better cooling, and you can overclock more.

If it has less radiator area, ten it's the opposite.

[KarathKasun]

3 minutes ago, Enderman said:

A loop only changes CPU temperatures, not performance.

The performance is affected by whether you overclock or not, assuming you have no throttling in the first place.

 

If the custom loop has more radiator area than the AIO you're comparing it to, then it has better cooling, and you can overclock more.

If it has less radiator area, ten it's the opposite.

If overclock more, you mean ~100mhz more, yes.  Its not a "good" investment unless you cant get a faster CPU/GPU model and you need the extra 0.5%.

[Enderman]

2 minutes ago, KarathKasun said:

If overclock more, you mean ~100mhz more, yes.  Its not a "good" investment unless you cant get a faster CPU/GPU model and you need the extra 0.5%.

How much you can overclock depends on the CPU you use, the silicon lottery and how lucky you got, how many radiators you use, what fans you use, etc etc etc.

The main point of a custom loop is looks and DIY fun, but if you use enough radiators you can get far better performance than any AIO that exists.

Obviously at a cost of time and money.

[JT241]

I must say as a first time poster it is incredibly fast how quickly and informatively this community replies.
Appreciate the help everyone who posted. I guess since I was mainly looking at water loops to help cool for future overclocking I'd be better off going for some heavy air cooling.

[KarathKasun]

6 minutes ago, Enderman said:

How much you can overclock depends on the CPU you use, the silicon lottery and how lucky you got, how many radiators you use, what fans you use, etc etc etc.

The main point of a custom loop is looks and DIY fun, but if you use enough radiators you can get far better performance than any AIO that exists.

Obviously at a cost of time and money.

System performance is not significantly impacted over big air or big CLCs, no matter how many radiators you use.

 

An extra sub 5% in clocks is not going to net a perceptible boost in games.

[Enderman]

4 minutes ago, KarathKasun said:

System performance is not significantly impacted over big air or big CLCs, no matter how many radiators you use.

Yes it is.

Maybe you should do a bit more research on custom loop watercooling.

 

There are hundreds of types of fans and radiators that allow you make configurations for silence, maximum performance, etc.

AIOs only really go up to 360mm, while custom loops you can have much more.

There are people getting very high overclocks on custom loops that wouldn't be possible with a single 360 aio or custom loop.

 

Obviously a custom loop can't match phase change, but it is the next best thing for temps and overclocking.

 

Maybe you're thinking of price/performance, not just performance.

[KarathKasun]

8 minutes ago, Enderman said:

Yes it is.

Maybe you should do a bit more research on custom loop watercooling.

 

There are hundreds of types of fans and radiators that allow you make configurations for silence, maximum performance, etc.

AIOs only really go up to 360mm, while custom loops you can have much more.

There are people getting very high overclocks on custom loops that wouldn't be possible with a single 360 aio or custom loop.

 

Obviously a custom loop can't match phase change, but it is the next best thing for temps and overclocking.

 

Maybe you're thinking of price/performance, not just performance.

AIO/big air + 9900k... 5-5.2ghz.  Full custom WC loop with ~$1000 in parts... 5.3ghz.  5.4ghz if you are lucky or buy a binned chip.

 

Big die/cluster CPUs can gain a bit more.

 

GPUs don't gain much considering where they top out on LN2.  Maybe 2300 core on a good NV sample.

 

The gains are typically sub 10% (much less for specific CPUs/GPUs) unless you are the type that buys stuff, overclocks it, and returns it when it doesnt hit your target.  Or you just get to put your hands on lots of hardware.

[Enderman]

7 minutes ago, KarathKasun said:

AIO/big air + 9900k... 5-5.2ghz.  Full custom WC loop with ~$1000 in parts... 5.3ghz.  5.4ghz if you are lucky or buy a binned chip.

 

Big die/cluster CPUs can gain a bit more.

 

GPUs don't gain much considering where they top out on LN2.  Maybe 2300 core on a good NV sample.

 

The gains are typically sub 10% (much less for specific CPUs/GPUs) unless you are the type that buys stuff, overclocks it, and returns it when it doesnt hit your target.  Or you just get to put your hands on lots of hardware.

Again, we're not talking about cost here.

OIt doesn't matter if it's $100 of parts or $1000 of parts.

What matters for cooling is radiators and fans.

 

Most people with custom loops do it for fun and looks, and only use a 360 or a bit more rad space.

Then there are overclocking entusiasts that use multiple 420 rads and high performance fans and easily get 5.5+ GHz

 

Clearly you're thinking about price performance instead of just performance.

FYI, those are two different things.

[KarathKasun]

7 minutes ago, Enderman said:

Again, we're not talking about cost here.

OIt doesn't matter if it's $100 of parts or $1000 of parts.

What matters for cooling is radiators and fans.

 

Most people with custom loops do it for fun and looks, and only use a 360 or a bit more rad space.

Then there are overclocking entusiasts that use multiple 420 rads and high performance fans and easily get 5.5+ GHz

 

Clearly you're thinking about price performance instead of just performance.

FYI, those are two different things.

You can get the same clocks with big air and really big fans (~300CFM).

 

Unless you are talking about perf/dB, which is also not just performance.

 

You tend to not see real big air OC's anymore due to the ease of just slapping a 360 CLC+decent fans on a setup, which can also get to the same clock range.

[Enderman]

7 minutes ago, KarathKasun said:

You can get the same clocks with big air and big fans (~300CFM).

 

Unless you are talking about perf/dB, which is also not just performance.

That's not how it works.

A custom loop allows for more cooling area to be used by simply expanding the loop, an air cooler or AIO does not.

So when using the same (absurd) 300cfm fans on the custom loop you would get better cooling performance than using those same fans on an air cooler.

 

Again, it all comes down to how much area you have.

Comparing a custom loop with slow fans to an air cooler with crazy fans is just a stupid comparison.

That's like saying "oh my GTX2060 perform better than your GTX2080 when you take the fans off of the 2080"

Like yeah obviously it does, because that's a useless comparison, you're purposefully crippling it.

[KarathKasun]

4 minutes ago, Enderman said:

That's not how it works.

A custom loop allows for more cooling area to be used by simply expanding the loop, an air cooler or AIO does not.

So when using the same (absurd) 300cfm fans on the custom loop you would get better cooling performance than using those same fans on an air cooler.

 

Again, it all comes down to how much area you have.

Comparing a custom loop with slow fans to an air cooler with crazy fans is just a stupid comparison.

That's like saying "oh my GTX2060 perform better than your GTX2080 when you take the fans off of the 2080"

Like yeah obviously it does, because that's a stupid comparison, you're purposefully crippling it.

You get to a point of diminishing returns due to blocks, pumps, TIM, and heat spreaders... even if you slap the same fans on the custom loop.

 

You can only cool the heat that you can extract from the silicon.

[Enderman]

11 minutes ago, KarathKasun said:

You get to a point of diminishing returns due to blocks, pumps, TIM, and heat spreaders... even if you slap the same fans on the custom loop.

 

You can only cool the heat that you can extract from the silicon.

Again, you can still get much higher heat dissipation with more radiator surface area.

Obviously there are other things that affect the cooling performance and yes there are diminishing returns, but a custom loop can still cool much more than an air cooler or AIO because of how you can have much more surface area.

 

I don't know why you think the laws of physics somehow don't apply here...

It's simple math and physics.

 

PS- people going for highest possible overclocks will use liquid metal, regardless of whether they use a heatsink, AIO, or custom loop.

 

PPS- with the powerful pumps used in custom loops the amount of blocks doesn't matter or impact performance, unless you're doing something dumb like 7 GPUs and 10 radiators. Even then, just use two pumps in series and there's no problem.

[KarathKasun]

9 minutes ago, Enderman said:

Again, you can still get much higher heat dissipation with more radiator surface area.

Obviously there are other things that affect the cooling performance and yes there are diminishing returns, but a custom loop can still cool much more than an air cooler or AIO because of how you can have much more surface area.

 

I don't know why you think the laws of physics somehow don't apply here...

It's simple math and physics.

 

PS- people going for highest possible overclocks will use liquid metal, regardless of whether they use a heatsink, AIO, or custom loop.

 

PPS- with the powerful pumps used in custom loops the amount of blocks doesn't matter or impact performance, unless you're doing something dumb like 7 GPUs and 10 radiators. Even then, just use two pumps in series and there's no problem.

Its not math and physics, its thermodynamics.

 

The lower the temp of your water/rad, the less heat you expel from your loop.  Once your water is ambient, there is no improving cooling.  It does not matter how much surface area and airflow you add.

 

At that point you have to increase flow rate of the coolant.  And you can not move beyond the pressure gradient that your block can withstand.

[Enderman]

4 minutes ago, KarathKasun said:

Its not math and physics, its thermodynamics.

 

The lower the temp of your water/rad, the less heat you expel from your loop.  Once your water is ambient, there is no improving cooling.  It does not matter how much surface area and airflow you add.

 

At that point you have to increase flow rate of the coolant.  And you can not move beyond the pressure gradient that your block can withstand.

Well if you knew anything about thermodynamics you would know that it is impossible for the water to get to ambient temperature.

[KarathKasun]

4 minutes ago, Enderman said:

Well if you knew anything about thermodynamics you would know that it is impossible for the water to get to ambient temperature.

Yes, because the approach to ambient is "infinite".  The closer you get the less heat you can remove, so you can only approach yet never attain.

 

This also applies to your block, and any other heat transfer system.

 

It just so happens that where you get to the point that water cooling is approaching ambient CPU temps is... just about the same as air.  Close enough that there is no advantage.

[Enderman]

6 minutes ago, KarathKasun said:

It just so happens that where you get to the point that water cooling is approaching ambient CPU temps is... just about the same as air.

You're not even close to correct.

Even with 5x 420mm ards you're still not 'close to ambient', and that is much higher cooling performance than any air or AIO cooler.

 

Your argument is that an air cooler gets the same CPU temps as a custom loop with multiple radiators, and that's literally a bunch of BS.

 

Go take a look at some watercooling videos from jayztwocents or singularitycomputers to see what kinds of temps and overclocks they get with large custom loops.

[Ttnuagmada]

You might be able to eek out a tad more from your CPU. IE lower voltages at your max clock, or maybe an extra 100mhz without too much more voltage.

 

Honestly, GPU gains are where water cooling really shows up.

[KarathKasun]

26 minutes ago, Enderman said:

You're not even close to correct.

Even with 5x 420mm ards you're still not 'close to ambient', and that is much higher cooling performance than any air or AIO cooler.

 

Your argument is that an air cooler gets the same CPU temps as a custom loop with multiple radiators, and that's literally a bunch of BS.

 

Go take a look at some watercooling videos from jayztwocents or singularitycomputers to see what kinds of temps and overclocks they get with large custom loops.

It seems you have not gone big air with big fans ever, I suggest you do some internet digging and look at what is possible with a properly setup air system.  It is not popular due to noise, and not many that do custom loops ever touch a good air benching system before writing them off.

 

The best comparison I have found is a Δ T of 40 vs 50, for an EKWB GPU block vs "high end air".  Going big fans over stock fans changes the Δ T of most heatsinks by ~10...

[Ttnuagmada]

31 minutes ago, Enderman said:

You're not even close to correct.

Even with 5x 420mm ards you're still not 'close to ambient', and that is much higher cooling performance than any air or AIO cooler.

 

Your argument is that an air cooler gets the same CPU temps as a custom loop with multiple radiators, and that's literally a bunch of BS.

 

Go take a look at some watercooling videos from jayztwocents or singularitycomputers to see what kinds of temps and overclocks they get with large custom loops.

To be fair, even a monster custom loop isn't going to give you too much more over something like an NH-D15. That's because the bottleneck of CPU heat removal is generally the heat transfer from the core to the cooling solution. It of course is going to depend on the CPU, but a typical non-HEDT CPU is only going to perform slightly better with a custom loop vs a good AIO or a NH-D14/D15.

 

Just to clarify, things change when we're talking about an HEDT chip.

[Enderman]

1 minute ago, KarathKasun said:

It seems you have not gone big air with big fans ever, I suggest you do some internet digging and look at what is possible with a properly setup air system.  It is not popular due to noise, and not many that do custom loops ever touch a good air benching system before writing them off.

 

The best comparison I have found is a Δ T of 40 vs 50, for an EKWB GPU block vs "high end air".  Going big fans over stock fans changes the Δ T of most heatsinks by ~10...

Air coolers are limited by the size of the heatsink, that's what determines the surface area for heat dissipation.

A custom loop is not limited because you can simply add more radiators.

And when you use the same fans on both, obviously the cooler with more surface area wins.

 

It's not hard...

same airflow + more surface area = lower temps.

You seem to have trouble understanding that.

[Enderman]

5 minutes ago, Ttnuagmada said:

To be fair, even a monster custom loop isn't going to give you too much more over something like an NH-D15. That's because the bottleneck of CPU heat removal is generally the heat transfer from the core to the cooling solution.

Which is why you use liquid metal if you're going for the highest possible overclock.

Regardless of whether you use a heatsink or custom loop or whatever.

 

That doesn't change the fact that more surface area = more heat dissipation.

 

 

I never said anything about it being worth the money or effort for the performance improvement, that's up to the enthusiast.

The point is that a custom loop allows for more radiators which means more surface area, which means lower temperatures.

Go take a look at jayztwocents and singularitycomputers on youtube for the numbers.