Liquid Cooling Myths

[malkiewicz]

Did you include other factors such as humidity / air temps / system configuration / amount of fluid used / ECT to your calculations Jay? Just asking....

[jayhall0315]

Good question Malkiewicz. Here is the basic setup I use every time someone releases a new super duper block claiming to beat all others which I am curious enough to buy myself:

Simple loop (open air, not in a case) in our back lab room that stays at 27 deg C, with a trusty Asus P6T Version 2 board and an Intel i7-980x processor at 4.4 GHz. The loop consists of the following which never changes - Swiftech MCP655 vario on setting 5 (4800 rpms) -> one foot of 1/2 inch ID, 3/4 inch OD clearflex tubing -> old but trusty Thermochill PA 120.2 with two Scythe S-Flex E fans blowing at 1000 rpms -> one foot of clearflex tubing "" -> 980x CPU with block custom clamped to maintain same even pressure (in Pa) -> one foot of clearflex tubing "" -> Swiftech micro-res filled to about 10mm clear of the top and fully bled -> one foot foot of clearflex "" back to pump. The TIM we always use is Arctic Silver 5. The mix we use is always 100 percent distilled water with 5 drops of algae biocide from Pets Smart. We use three software and two hardware designs to monitor temps. I must admit though, I fooled with this setup much more when I was back in grad school. Now that I am working as a PI, I usually only test a block on this setup perhaps twice a year and only for personal curiosity.

Blocks I have purchased and put thru this test out of my own curiosity : Apogee XT, Apogee XT rev 2, EK Supreme HF (with three different plates), the XSPC Rasa (not tested), the Koolance 360, the Apogee GTZ, and several older blocks going back to the Maze 4. Whenever I do testing, I generally find agreement or pretty close agreement Martinm when our results cover the same areas. I often do not find agreement with testing done by Skinnee.

I invite you to test this for yourself readers and you will see that the ten year old Maze 4 block is not that far off. I dont own the Storm block, so cant comment there Stephen.

[stephenswiftech]

Thanks for the reply Stephen. No disrespect, but I cant replicate your results. My ole Danger Den Maze 4, is about 3 to 4 deg C back from the Apogee XT Rev 2 for the socket 1366 testing I did in August. Dont get me wrong, I like Swiftech. In fact, I have your blocks on my home system (bc I like the easy mounting system and the MCW 82 universal gpu blocks which save me money). I also use your stuff in my lab, but I sure dont see any 8 deg C or larger difference between early blocks and the Apogee XT rev 2.

If you are the block designer, or even better, a real university educated engineer, then you know the thermodynamic limits of 2D gaped array heat exchange. You may not be solving it by hand (hehe), and using some cfd software for quick estimates, but the heat equation and Navier-Stokes are very specific (although 3D N-S is of course extremely chaotic bounded by initial condition). You can cut this groove, or add this inlet, but only so much water can be forced thru for a given pressure with a given Reynolds turbulence. Of course if you add in 2, 3, ...10 more inlets, you will see marginal Delta T drop, but 2D array is limited. The current crop of Apogee XT, EK super duper, and Koolance Super Star Destroyer are rounding the bend on what you will get from a 2D finned array. I suppose if you add in 4 inlets and 4 outlets with finned gaps to match each channel and four pumps to push it, etc...then you will see real gains. But, I am restricting the discussion for my part to blocks which have one 1/2 inch inlet and outlet.

There is a way forward with copper blocks, but not in the manner in which most companies are copying each other now (or making very incremental improvements). Unfortunately, the patent process is in the final leg so to speak, so I would not like to reveal it just quite yet. (nor will I be making it myself once the patent is finalized). It is common idea that comes from submarine propellor design and N-S optimization is all I care to reveal now.

I haven't had the chance to test the Maze4. If you tell me you had the equipment to make a true apple to apple test, I will not doubt your numbers. Actually I wouldn't be surprise. I've seen some surprising results over the past few years.

But the thing is, we are (unfortunately - because CFD and hand-on fluid analysis were lots of fun) no longer in the same ball game anymore. And to be honest, it's been like that for a good 4 years. Today manufacturers can obviously all produce high density base plates and the raw copper to coolant thermal conductivity is pretty high for everyone. Actually, it's that high that is very difficult to even measure these differences using even good lab equipment.

Now the real battle for performance happens around the mechanical contact. Everyone has heard how not flat CPU's IHS are and everyone has at least heard the word "bow". There are actually not so many researches that have been done on the thermal aspect of cyl/cyl mechanical contacts. At some point I even considered rerolling my PhD in this area but i've got way too much work here , ahah. More seriously, your assumptions are correct, all improvements made around the fluid dynamics in modern water blocks will be incremental. But that doesn't mean it stops right here. That's quite the opposite, every time a new CPU or socket is released it usually changes everything, and that's why we need to keep researching this so we can keep producing products that are designed for these modern CPUs.

with today's technology if you see someone claiming more than incremental temperature gains because of a slight change in the hydraulic of a block then you're probably looking at a myth and whatever gain it was, it was probably due to something else.


edit: a couple more things. Socket 775 really changed things with how the IHS is/was clamped. Later, dual cores (and then especially double separate dies) changed things also on how/where CPU temperature was read. LGA1366 and the large die also changed a bunch of other things... Now LGA2011 is coming, double clamping and an even larger IHS... fun fun stay tuned =D

[jayhall0315]

You might be a great contact person in the future Stephen. If you could, send me your contact details by PM and I will stash em for future reference. What I will be curious to see is how the Apogee HD performs relative to the Apogee XT Rev 2 with one inlet and one outlet on the 980x at 4.4 GHz. Also, contrary to what Skinnee published, I actually show the Apogee XT neck and neck with the EK Supreme HF and I do not get the 1.8 deg C difference he notes (on average).

[stephenswiftech]

Most of the time differences from a user to another (or from a tester to another) when comparing two identical products will be due to either quality of mounts or simply because they don't use the exact same CPU (different IHS concavity/convexity, etc). Which is one of the reasons we typically test different CPUs. I'll send you my contact info in a PM.

[NaeKuh]

The number of soap boxes dropped?

lol couldn't resist

Thats the problem!!! instead of soap boxes its a wooden bucket.

do you know how easy those are to break? skinnee wouldnt need a hammer for those... he just needs to kick them.

The longer you talk to him, the more you realize there are always silent/un-stated implications in his argument.

:O

ummm... no... i assume.. umm.... u have telepathy to understand the 75% i leave out.

[MrBean]

3. You need a 480 rad to cool anything newer than a P4.

I cannot agree more with you, one of a few using their grey matter.

In actual fact, even a 480mm rad is overkill. But, tell that to some of the Boyz around here, and you'll get shot. Let's hope they're not reps-in-disguise I think it is a case of people not understanding the difference between their perceptions and reality, when it comes to the cooling requirements of modern CPU's.

Enough of my rant, but I have to add, yes, this is the biggest myth of all time (the too-big-rad-myth)

ps: Mmm, like your ideas wrt a new design, and referencing ship/sub propellers - all I can think of you're reference has to do with "cavitation" effect of modern props, or something in that line...you know, causing an area of low-pressure, resulting in water boiling at much lower temp, even close to room-temp, resulting in "phase-cooling" effect, well, in a normal waterblock......good luck, about time for a fresh approach....

To close for comfort, I hope not :flowers:

[Alexandr0s]

In actual fact, even a 480mm rad is overkill. But, tell that to some of the Boyz around here, and you'll get shot. Let's hope they're not reps-in-disguise I think it is a case of people not understanding the difference between their perceptions and reality, when it comes to the cooling requirements of modern CPU's.

And of course people like Naekuh who want a <1C delta !

[MrBean]

And of course people like Naekuh who want a <1C delta

Yeah, then there's Naekuh But, he's on a planet of his own, hehe. I am talkin the general populace here, that excludes him.

Soz Naekuh, had to.....

[Kallenator]

I cannot agree more with you, one of a few using their grey matter.

In actual fact, even a 480mm rad is overkill. But, tell that to some of the Boyz around here, and you'll get shot. Let's hope they're not reps-in-disguise I think it is a case of people not understanding the difference between their perceptions and reality, when it comes to the cooling requirements of modern CPU's.

I actually like to think of it this way:
X amount of heat, needs Y amount of radiators and Z amount of fan speed.
The more Y you have compared to X the less Z you need to get acceptable temps with a given power consumption. And when radiators are as cheap as they are and since most of us are using fairly big cases, one might as well use the space available.
And the reason I think like this is: I like the potential of liquid cooling system, but I would like to take advantage of the possibility to have a really quiet system as well.

[MrBean]

I actually like the think of it this way:
X amount of heat, needs Y amount of radiators and Z amount of fan speed.
The more Y you have compared to X the less Z you need to get acceptable temps with a given power consumption. And when radiators are as cheap as they are and since most of us are using fairly big cases, one might as well use the space available.
And the reason I think like this is: I like the potential of liquid cooling system, but I would like to take advantage of the possibility to have a really quiet system as well.

How much time does your PC actually spent under 100% full load?

There are very good automatic controllers on the market, which you can set your own control curves, to assist with taking care of the noise-issues, if any....

But, I knew someone was going to come with this argument, I mean, you can have a quiet system, with 1/2 to a 1/3 of what most guys are 'claiming' you need to 'cool' such and such a system. Again, it's all in perceptions...

But, this takes it off-topic, and a discussion for another thread.

[MrBean]

I actually like the think of it this way:
X amount of heat, needs Y amount of radiators and Z amount of fan speed.
The more Y you have compared to X the less Z you need to get acceptable temps with a given power consumption. And when radiators are as cheap as they are and since most of us are using fairly big cases, one might as well use the space available.
And the reason I think like this is: I like the potential of liquid cooling system, but I would like to take advantage of the possibility to have a really quiet system as well.

How much time does your PC actually spent under 100% full load?

There are very good automatic controllers on the market, which you can set your own control curves, to assist with taking care of the noise-issues, if any....

But, I knew someone was going to come with this argument, I mean, you can have a quiet system, with 1/2 to a 1/3 of what most guys are 'claiming' you need to 'cool' such and such a system. Again, it's all in perceptions...

But, this takes it off-topic, and a discussion for another thread.

[Church]

stephenswiftech: If lately it all became mostly about mechanical contact quality to differentiate cooling performance of recent blocks @ recent CPUs, i wonder what can be done to make it best possible. Seeing top air coolers weighting now about kilogram, seeing performance improvements and lessening of TIM influence with much more pressure ('great contact' mounts @skinneelab TIM tests), imho safety margin for maximum mount pressure of water blocks can be decreased and vendors should squeeze out more performance by using much more pressure then on average it is now.

[Kallenator]

How much time does your PC actually spent under 100% full load?

There are very good automatic controllers on the market, which you can set your own control curves, to assist with taking care of the noise-issues, if any....

But, I knew someone was going to come with this argument, I mean, you can have a quiet system, with 1/2 to a 1/3 of what most guys are 'claiming' you need to 'cool' such and such a system. Again, it's all in perceptions...

I would think my CPU is around 150W+ at the moment (50% more freq and 1.45vcore, 95W * 1.5 ++). So for a 3x 120mm radiator, yes, I agree with you even for a silent system its enough.
Used to have a 4870x2 on it though, so I didn't choose that RAD just for having it on the CPU alone.
There is also the question about what I consider acceptable noise level and I really doubt that when I get the extra 6950 and put both under water that the 360 will be enough with noise well under 30dB during load. But I might be wrong still, I am going to try it out though when the parts arrive!

And for the fun of it, let's say you already have an 800D, then it would be cheaper for you to buy more rads and more fans and just keep them at your acceptable level of noise rather than buying for example an "Koolance TMS-200" to control them. ^^
(Not an entirely serious argument, just pulling your leg a little Bruce)

[Annoyingmouse]

How much time does your PC actually spent under 100% full load?

There are very good automatic controllers on the market, which you can set your own control curves, to assist with taking care of the noise-issues, if any....

But, I knew someone was going to come with this argument, I mean, you can have a quiet system, with 1/2 to a 1/3 of what most guys are 'claiming' you need to 'cool' such and such a system. Again, it's all in perceptions...

But, this takes it off-topic, and a discussion for another thread.

That's a great point as well as, what is realistic for 'full load'. If I run prime95 and UniEngine's Heaven demo (vsinc off) side by side, my kill-a-watt tells me I'm using well over 300 watts, but no real world usage including video gaming with folding@home running in the background even comes close to that.

[Tom128]

Home Depot Brand Tubing is great.. <--- :\ sure if u like it to cloud in weeks...

Just throwing it out there since we are talking about myths and being informed and what-not. From a performance standpoint that's kind of irrelevant. I've been water cooling for a few years now and have almost always used tubing from Lowes since I can walk across the street to get it. However, I also use black tubing most of the time so clouding isn't a problem. The biggest downside to Home Depot/Lowes type tubing is that it is not as good at bending without pinching.

[c.ruel]

I don't know if it's just a myth or me, but for about the first year of my getting into watercooling I believed that watercooling would dump way less heat into the room because the cpu temps were lower.

if it is less, it's not much.

That doesn't make any sense at all. Your cpu no matter how you cool it produces a fixed amount of heat. If you cool it with aircooling your room will heat up slower because the heat will be transferred to the room less efficiently. If you cool your system with water the room will heat up faster but to the same eventual temperature because the cooling system you have is more efficient.

If anything under most circumstances your room will get hotter will watercooling because most people will up their overclock and voltage making the cpu produce more heat, that heat has to go somewhere ... the fans blow it into the room ... blowing the heat into the room.

It's like naekuh said, you can't trick physics.

[fr0wn3r]

That doesn't make any sense at all. Your cpu no matter how you cool it produces a fixed amount of heat. If you cool it with aircooling your room will heat up slower because the heat will be transferred to the room less efficiently. If you cool your system with water the room will heat up faster but to the same eventual temperature because the cooling system you have is more efficient.

If anything under most circumstances your room will get hotter will watercooling because most people will up their overclock and voltage making the cpu produce more heat, that heat has to go somewhere ... the fans blow it into the room ... blowing the heat into the room.

It's like naekuh said, you can't trick physics.

Well said!

If you want to warm your room in the winter OC your HW and watercool it. It's very efficient in dissipating heat in the surroundings.

[stephenswiftech]

That doesn't make any sense at all. Your cpu no matter how you cool it produces a fixed amount of heat. If you cool it with aircooling your room will heat up slower because the heat will be transferred to the room less efficiently. If you cool your system with water the room will heat up faster but to the same eventual temperature because the cooling system you have is more efficient.

If anything under most circumstances your room will get hotter will watercooling because most people will up their overclock and voltage making the cpu produce more heat, that heat has to go somewhere ... the fans blow it into the room ... blowing the heat into the room.

It's like naekuh said, you can't trick physics.

Like it's been said before there is a difference between myth and typically immeasurable physics. Behind this one myth there is an immeasurable fact which invalidates this myth.

As the CPU temperature decreases, its impedance (~ resistance) decreases which increases the current (voltage behind regulated), as the current incrementally increases, the CPU heat load also incrementally increases.

As you cool down your CPU, you also increase its heat rejection into the environment. Talking incremental numbers here (to give you an idea: <1% heat increase for ~ 10C drop)

[tiborrr]

True what Stephen said. I've did a quick test when warming up the FX-8150 from 100°C. Prime95 @ -100°C = 160W draw at the socket, @+20°C = 190W.

[CrazyNutz]

Your cpu no matter how you cool it produces a fixed amount of heat.

Actually it does not. It not linear because silicon becomes more resistive as the temperature rises.
Hotter silicon will have higher resistance, the higher resistance will add heat.

[defect9]

That doesn't make any sense at all. Your cpu no matter how you cool it produces a fixed amount of heat. If you cool it with aircooling your room will heat up slower because the heat will be transferred to the room less efficiently. If you cool your system with water the room will heat up faster but to the same eventual temperature because the cooling system you have is more efficient.

If anything under most circumstances your room will get hotter will watercooling because most people will up their overclock and voltage making the cpu produce more heat, that heat has to go somewhere ... the fans blow it into the room ... blowing the heat into the room.

It's like naekuh said, you can't trick physics.

well, yeah, i know that now.

but back then I wasn't that versed in any physics. so it made total sense.

[stephenswiftech]

Actually it does not. It not linear because silicon becomes more resistive as the temperature rises.
Hotter silicon will have higher resistance, the higher resistance will add heat.

You got it backward. Resistance does rise with temperature but current decreases as the resistance goes up (fixed voltage). And lower current = lower heat output.

A cool CPU will generate more heat than a hot CPU (assuming the same clock, running the same application, blablabla)

[relttem]

You got it backward. Resistance does rise with temperature but current decreases as the resistance goes up (fixed voltage). And lower current = lower heat output.

A cool CPU will generate more heat than a hot CPU (assuming the same clock, running the same application, blablabla)

It is interesting to think about in terms of coolant; if coolant A cools better than coolant B would that mean that there is more heat being released by the cooler chip? So, the CPU temp would actually be higher for the better coolant..wacky. Or, would the better coolant be able to take care of the extra heat that is being generated by the cooler chip, thus keeping the CPU temp the same (relatively speaking)

[CrazyNutz]

You got it backward. Resistance does rise with temperature but current decreases as the resistance goes up (fixed voltage). And lower current = lower heat output.

A cool CPU will generate more heat than a hot CPU (assuming the same clock, running the same application, blablabla)

The current does not decrease (perhaps it can a bit) it is rather wasted as heat. As the CPU heats up you'll see the current demand rise as well.

I've did a quick test when warming up the FX-8150 from 100°C. Prime95 @ -100°C = 160W draw at the socket, @+20°C = 190W.

Tiborrr's test shows this well

where did the extra 30w go when the CPU was warmer?, did the cpu convert the 30w to more calculations? I think the 30w got converted to Heat!