Motherboard water-cooling question

[Rudy Toody]

I posted this at Motherboards to find out about water-cooling options for my new rig. I welcome your comments, suggestions, and water-cooling jokes.

 

[dma0991]

Just place a fan blowing towards the stock passive heatsink would be sufficient. If the CPU is watercooled, you wouldn't get the usual draft from air cooling to the motherboard heatsink.

The cheapest way would be to just have a fan blowing at it. Watercooling it will be better but that is only if you're willing to spend on it.

 

[BonzaiDuck]

So far, I've been an interested bystander on the matter of AiO and custom-water solutions. If "I'm gonna do it soon," I don't want it to be some half-assed half-successful project.

There are "mindsets" in the various cooling "camps" which can obscure the options.

Once you commit to water-cooling, you may actually conclude that (a) you don't need a lot of airflow in the case, or (b) everything must be "water-cooled."

But water-cooling requires "air-cooling" to be effective without any sort of external passive (fanless) solution that works to any degree of satisfaction. And about all the air-cooling solutions in front of us these days includes a certain amount of "liquid cooling," since those solutions generally depend on heatpipes or vapor-chambers.

The more complicated the water-cooled rig, the more fittings and hoses, the greater the possibilities of some leak or failure. What those possibilities are -- we really don't know. They may not all be that great.

Then consider the expense of extra parts. If you can cool motherboard components more than adequately with airflow -- which is cheaper -- why complicate the picture with additional water-blocks, hoses and connections?

There's no reason to insist that the entire system use one cooling strategy exclusively.

For instance, it's easy to imagine someone with an ASUS Sabertooth motherboard choosing "custom-water." But the plastic motherboard plate on those boards has a purpose: to air-cool the motherboard components. The plate includes one or more fans which force air into the narrow space underneath the duct-plate.

 

[Rudy Toody]

I already have all the watercooling stuff. The initial plan is to cool the cpu, memory, nb, and vrm. Later on I will do some gpus (when I can afford them.) The only waterblock I don't have is one for the vrm.

So, I need to know if there is a vrm waterblock for the ASRock. I already know EK has one for the ASUS.

If I can't watercool the ASRock, I will have to go with the ASUS and EK. I prefer ASRock because I have never had any problems with them.

 

[BonzaiDuck]

I already have all the watercooling stuff. The initial plan is to cool the cpu, memory, nb, and vrm. Later on I will do some gpus (when I can afford them.) The only waterblock I don't have is one for the vrm.

So, I need to know if there is a vrm waterblock for the ASRock. I already know EK has one for the ASUS.

If I can't watercool the ASRock, I will have to go with the ASUS and EK. I prefer ASRock because I have never had any problems with them.

I understand your position. I should avoid suggesting that you're taking this stuff to an extreme: the hobby includes all sorts of preferences, obsessions, perfectionist strivings -- and there's room for all of it. That way, a budget builder and a Maximum-PC-Dream-Machine Replicator can share useful information.

If anything bothers me, it is the investment in parts and features that I may have little use for three years in the future. But that's MY business, and what others choose to do is THEIR business.

But whether or not you already have the parts, the only thing that would make water-cooled NB and VRM more than a marginal improvement in a diminishing-return scenario would be "extreme" overclocking. And you could also cool those parts down with a little attention to airflow, 40mm fans and some motherboard ducting. How much better water-cooling would be or how it might pay off, maybe you know better but I am "dubious."

AsRock is "coming into its own." No doubt about it. But ASUS is no slacker. The last time I "had trouble" with an ASUS board, it was 2007, the board was a 680i (NVidia) board, and I should've paid more attention to the reviews before I bought it. It's still running though: in my home-server system.

 

[PhIlLy ChEeSe]

Cooling memory is a waste of time IMHO, if the parts are new and un used I'd list them on Flee bay. Water wont do anything air cant do................

 

[BonzaiDuck]

Cooling memory is a waste of time IMHO, if the parts are new and un used I'd list them on Flee bay. Water wont do anything air cant do................

I learned from my recent Seasonic PSU failure that parts can slowly cook to death during lengthy sleep-states. If the fan can't come on, the process is stressful to PSU parts. I have left my AVR on for months, and the heat dissipation has given me a side-project of installing a fan in the top of my component cabinet, when it is totally unobstructed from the rear, with shutter doors -- shutters wide for line-of-site to handheld remote.

Therefore, for computer -- sleep after usage ends in 1 or 2 hours, hibernate after 3 or 4. For the AVR, leave it in sleep pass-through state, switched on for special events or broadcasts.

I'd like to think that a failing PSU can cause data-corruption. It would be the nature of the failure that determines any damage to components. At least one of those scenarios could include RAM.

With some exotic cooling systems, you can maintain a tight lid on rise from idle temperatures to load. I think Guskline said 40C, and he had an MO-RA3 and a 360mm radiator at the front of his midtower case. Bong-cooling believers can hold the temperature range to several degrees below ambient, to just above it.

But the parts are made to assume certain operating conditions and sustain them. If temperatures fall within the expected range published for the Intel product or other component, you might not expect any improvements in performance of consequence by more efficient and vastly more effective cooling systems. But you may have expended a pile of change in the investment. Then, you would likely grow to have a religious attendance to routine maintenance, or otherwise assuring that the water-cooling system keeps working as it should.

The main thing to remember is that enthusiast machines have for a long time been a kind of art form. Maybe it's reasonably likely that subsequent use of the watercooling parts is possible in newer builds or upgrades. Or simply a swap of motherboard, CPU, RAM and video card(s). Perhaps -- the occasional change-up to a better water-block or radiator.

So as AigoMorla has already said -- it is an actual "commitment."

 

[KaRLiToS]

Cooling memory is a waste of time IMHO, if the parts are new and un used I'd list them on Flee bay. Water wont do anything air cant do................

It will but it will only bring very small marginal gains. Water will dissipate the heat outside of the case... while air will cool it away from the sticks but the heat will stay inside the case.

But I agree with you, I'm a big watercooling guy and memory is about the only thing I will not watercool along with the hard disk drives.

 

[aigomorla]

So, I need to know if there is a vrm waterblock for the ASRock. I already know EK has one for the ASUS.

try koolance.

They have plates which u can get to approximate size.

http://koolance.com/index.php?route=product/category&path=29_148_108
+
http://koolance.com/mvr-100-motherboard-vreg-water-block

Make sure u read the discription tho on the Mvr-100... it only works with 1 cold plate.
So if your mosfet area is larger then the Mvr, you will need the 140mm plate.
If its smaller you will need the smaller mosfet block:
http://koolance.com/mvr-40-motherboard-vreg-water-block

and then assemble the block.

It would look simular to my Mosfet block...

when you combine them together you get this:

which then u apply on your board to look like my mosfet block:

Just make sure you measure out the exact length you require in aspect to the cold plate needed.

My thoughts on memory water cooling is a COMPLETE waste.
Your best off getting a memory cooling fan like this:
http://www.newegg.com/Product/Produc...-003-_-Product

The only time i would even consider watercooling ram would be on FB-ECC-Registered, and even then someone would need to pay me to do it.
The plumbing involved + parts + unflexability to swap modules out when required... = MASSIVE HEADACHE

 

[BonzaiDuck]

It will but it will only bring very small marginal gains. Water will dissipate the heat outside of the case... while air will cool it away from the sticks but the heat will stay inside the case.

But I agree with you, I'm a big watercooling guy and memory is about the only thing I will not watercool along with the hard disk drives.

You can get a heat-pipe cooled system to expel the heat immediately outside the case. Noobs and mainstreamers won't bother with the tedium, though. And the tedium cuts both ways: look at how Lepton87's 5820K project is shaking out.

I have a flexible view of all this.

I have to ask whether some watercooling enthusiasts ever had some dream about reducing airflow, or making their rigs look like art-museum showcases.

You could watercool every heat-generating part inside a system, but the complexity and maintenance problems would multiply.

With "bong-enhancement," you can watercool to get sub-ambient temperatures, but even "just above ambient" would be an excellent result.

With ample radiator capacity, your overclocked stress-load temperature under IBT or LinX in the 75F-ambient-range can be held to 40C.

But the expected operating temperature of these processors is more forgiving, I would think. So with air and AiO options, maybe requiring some DIY-customized attentions, high-CFM-noise-reduction remedies and so on, you might give up a hundred Mhz or two in clocks, but you might reach a limit in voltage defining a self-imposed "comfort range."

You might narrow the disparity with some AiO models. To the extreme, you could invest in a pair of larger radiators, dual-pump and reservoir.

Somewhere on that end of the options, you have a luxury of pushing voltage as well as the risk of pushing it.

So it's also a question of whether or not the few Mhz really matter to you, and how often you plan to turn over your system parts for upgrade. Or whether you envision redeployment of certain parts, or a maintenance-schedule associated with them.

 

[Lepton87]

What's a bong-enhancement? What's the point of dual radiator? My Mo-RA3 doesn't even heat up enough to notice it. In cars radiators of this size dissipate tens of KW, I can't imagine any system using that much. My radiator under current load works fine even without fans, I checked it.

 

[aigomorla]

What's a bong-enhancement? What's the point of dual radiator? My Mo-RA3 doesn't even heat up enough to notice it. In cars radiators of this size dissipate tens of KW, I can't imagine any system using that much. My radiator under current load works fine even without fans, I checked it.

Bong cooler is based on the principles of evaporation.
It can go lower then ambient by a few C.
However its has lots of water loss, and can cause a spike in humidity.

Dual radiators are used to split the heat capacity up so you can mount them in different locations.
like have one up front, and then another on top.

Some people also sandwich them like this:

The only downside to it, is that unless u have powerful fans to move air though the restriction, the performance will not be so great.
If you do have high static fans tho, the performance will most likely be better then your MORA due to the flat tubing structure of that rad with 2/3rds with surface area.

Edited cuz i felt i was sort of derailing the OP's thread...

 

[BonzaiDuck]

What's a bong-enhancement? What's the point of dual radiator? My Mo-RA3 doesn't even heat up enough to notice it. In cars radiators of this size dissipate tens of KW, I can't imagine any system using that much. My radiator under current load works fine even without fans, I checked it.

Bong cooler. Volenti cooler.

They're called "bong" coolers because the evaporation tower with a "Y-joint" that blows air into the chamber looks like a water-cooled marijuana or hashish pipe:

http://wc101.com/guides/bongs

http://www.overclock.net/t/511330/horizontal-volenti-cooler-56k-warning

http://www.overclock.net/t/1374568/volenti-cooler-build

http://www.overclock.net/t/557313/rays-bong-build-how-to-56k-warning-come-see

DUAL RADIATORS

You link two radiators in series. The thread on the MO-RA3 that I'd started before you posted your own seems to suggest that a dual-pump configuration is better in the dual-radiator scenario. And the thought implicit in your third sentence eliminates any need for explanation: scalable increase in heat dissipation means lower temperatures.

But if you're either stating an argument or answering a question, that had also been my own thought: there's only so much cooling capacity needed for this or that processor -- with this or that TDP and other factors -- to function entirely within an expected range of operation for stock settings under an overclock, with common-sense case-design.

That's what I meant by an "optimum:" what difference would it make for overclocking potential and regular operation if a processor would heat to 55C under severe load, or never exceeded room-ambient with identical stress?

It would only make a practical difference if it would allow you to significantly reduce voltage for the same or higher clock-speed.

Do you remember your MO_RA3 load temperature when you had the CPU clocked to 4.4?

See, I'm no less interested in your MO-RA3 than I am in the bongs and Volenti's. The latter seem to be too much trouble for too much cooling and too little mobility. But I can even see ways to re-build and modify a mid-tower case which would make the MO-RA3 integral to that case so to avoid obstructing maintenance or access. On the other hand, that's a lot of work . . . too . . .

And actually -- AigoMorla had stated his same view on the MO-RA3 in posts to my earlier thread on it.

 

[Lepton87]

Do you remember your MO_RA3 load temperature when you had the CPU clocked to 4.4?

.

Those Valenti coolers seem like air humidifiers, I'll pass. The temperature never exceeded 80C and that's with 1.3V with a power virus, under normal load it is about 15C lower.

The only downside to it, is that unless u have powerful fans to move air though the restriction, the performance will not be so great.

Seems like way too much noise, my fans are practically inaudible.

 

[KaRLiToS]

Those Valenti coolers seem like air humidifiers, I'll pass. The temperature never exceeded 80C and that's with 1.3V with a power virus, under normal load it is about 15C lower.

Seems like way too much noise, my fans are practically inaudible.

Lepton, to calculate the Delta T (Or the efficiency of the radiator) you need to have some data. Touching it or comparing CPU temperature under load is not the correct way.

You need to have a air sensor inside your room and you need to have a water temperature sensor.

If your Ambient is 25'C in the room and your water temperature are 28'C. Then you have a delta T of 3'C. The water temperature is really important for that.

That is why I installed a couple of sensors in my rig. I haver two sensors in the water loop, one air sensor outside my Computer in my room, and I also have one sensor inside the case to calculate the temperature of the air inside the case.

But in your case lepton, you went overkill to have a quiet cooling system, which is pretty good. A 1080mm radiator with low rpm fans is perfect for this goal. You succeeded in doing a quiet system, having a single 360mm wouldn't achieve the same noise level.

 

[Lepton87]

The water is almost ambient, at least I can't measure any increase above ambient with my thermometer. I used the thermometer from my aquarium so it's not super precise.

 

[KaRLiToS]

The water is almost ambient, at least I can't measure any increase above ambient with my thermometer.

Then you have a pretty good watercooling loop. Remember you asked me previously if my loop was overkill:

all my fans are set to medium and when I'm playing intensive games, my ambient is around 25'c and the water temps can vary around 30'C and this is with my 18x120mm of radiators. Delta T of 5'C

So my loop is not that overkill considering the amount of blocks I have inside it. Yes I could live with a Delta T of 15'C but I wanted efficiency and low noise level.

 

[Lepton87]

Then you have a pretty good watercooling loop. Remember you asked me previously if my loop was overkill:

all my fans are set to medium and when I'm playing intensive games, my ambient is around 25'c and the water temps can vary around 30'C and this is with my 18x120mm of radiators. Delta T of 5'C

So my loop is not that overkill considering the amount of blocks I have inside it. Yes I could live with a Delta T of 15'C but I wanted efficiency and low noise level.

I'd say it's completely overkill for a one block and a quite modest OC but that's what I expected because I want to add 2 or 3 cards to it.

 

[KaRLiToS]

I'd say it's completely overkill for a one block and a quite modest OC but that's what I expected because I want to add 2 or 3 cards to it.

Here is an example of intensive gaming on my rig ---- VIDEO

During those period, using 4 x Cards at full load and the delta T can raise to 5'C which I consider pretty good.

 

[Lepton87]

Here is an example of intensive gaming on my rig ---- VIDEO

During those period, using 4 x Cards at full load and the delta T can raise to 5'C which I consider pretty good.

Wow, you have probably in excess of 5 times more heat coming from your loop than mine.
ps. I somehow don't like multiple monitors, the bezels are killing the experience for me I'd prefer a single 4K, but nothing can handle 3x4K right now so that's a no go for you.

 

[BonzaiDuck]

The only thing I can guess: the temperatures on some of the graphics cards will be lower if the game doesn't consume 100% of them all. But with the triple-monitor setup and resolution, it's really going to warm up those puppies.

I once had GRID installed, but somehow it became unfunctional and I purged it. Then I got Assetto-Corsa Simulator. Those two simulators were made by the same software design team -- unless there were just "commonly shared libraries," but the tracks and other features -- like the green-arrow path -- is identical.