Here we have a fitting waterblock for DFI NF4 !!!

[situman]

it looks to me the tubing will touch the card.

[@njo]

very nice.. but i would like to see a waterblock in the chipset (NB) and two boards in SLI...

[HARDCORECLOCKER]

The Waterblock shown in first post now has been modified by the developer so that it will fit on DFI NF4 together with two SLI cards and NV40 cooler.

[stu_allen]

Radical_53: About your euro perspective. The euro blocks work well at low flow, BUT BUT BUT... they need HUGE pumping power to get even little amounts of flow.. for instance the nexxos xp wont even get over one gpm with an IWAKI!! The american blocks give you = to or better performance and you get alot more flow which means that you get better cooling in the rad and the other blocks in the system will be getting more flow as well... All in all a better performing system.

You dont need that flow though.
I use a german 8mm system, with 4 drive coolers, 2 gpu coolers, nb cooler, and cpu cooler plus res and 3x120 rad. Flow is 0.7l/min, with the flow sensor as last thing in the loop apart from the gpu's. With both gpus clocked over ultra speeds, adn a fx-55 at 2760 i dont go over 46deg under load. Well within limits, and stable. Who cares if it could be 2-3 deg cooler?

[HARDCORECLOCKER]

You dont need that flow though.
I use a german 8mm system, with 4 drive coolers, 2 gpu coolers, nb cooler, and cpu cooler plus res and 3x120 rad. Flow is 0.7l/min, with the flow sensor as last thing in the loop apart from the gpu's. With both gpus clocked over ultra speeds, adn a fx-55 at 2760 i dont go over 46deg under load. Well within limits, and stable. Who cares if it could be 2-3 deg cooler?

I can confirm this..........

[jinu117]

46 degree on? How are you measuring this? If it is from mobo diode, that does't mean a thing to me. What temp are you reading? If it is GPU temp, wow!!! that's high for H2O.

[Torin]

You dont need that flow though.
I use a german 8mm system, with 4 drive coolers, 2 gpu coolers, nb cooler, and cpu cooler plus res and 3x120 rad. Flow is 0.7l/min, with the flow sensor as last thing in the loop apart from the gpu's. With both gpus clocked over ultra speeds, adn a fx-55 at 2760 i dont go over 46deg under load. Well within limits, and stable. Who cares if it could be 2-3 deg cooler?

I don't know what GPU clocks you are getting, but an FX-55 @ 2760 on water and 46C under load is plenty reason to want more flow. You could be doing better with the stock air cooling. When stock air is better than your water cooling, you can't say you don't need the flow. In your case, flow is not the only thing it seems you need. I'd imagine you are in it for looks/noise, not performance?

[Radical_53]

@jinu: Any temp you don't measure yourself with your own sensors can't be compared the real way.
For the GPU, I've seen differences especially on 6800 series, of around 20°C comparing different cards. CPU temps could only roughly be compared if some people use the same board with the same bios.
Example: My 6800 GT got up to 90°C under load on air, only 47° load with water. Quite a difference. Still, some people reported ~60°C under load on air. Can't be compared, right?

@torin: 46°, 2.8GHz, stock cooling? Of course... 30° load with water cooling, eh? Stop kidding around.
There's no "magic" with water cooling, just stuff that can be measured. It has been compared more than once, and the result was more than clear.

[jinu117]

Excuse me but that is what I asked. If your CPU block temp is reading 47c on water cooling, it is tad high for me. My FX-55 running 2.95ghz won't get near there :P GPU temp is even more drastic but it seems to me ur using 6800ultra so that is left to be judged (back of card temp reading always is good on this and mine gets to 33c with jamming 1.64v and 620/590 on X800XT), of course, my setup doesn't happen to be typical setup with 2 MCP-600 dual heatercore, 1/2" id and G4 + Silverprop HL. But to dismiss flow doesn't mean much in many block is... errrr... just bad assumption granted other things don't hold up. I turn one of the pump off and I see 3-4c increase temp on blocks (about 2-3c on water temp increase). There is no magic in water cooling. More flow will always net you better temp and in this case, with such restriction, you need all the flow you can get with chipset block to increase performance. I've had both NV6800 block and Maze4 GPU block on 6800 Ultra at one point and temp reading differed as much as 7c due to flow killer (nv6800) on MCP350 and BIM2. (exact same card) (Really, what a waste of money on that nv6800...)
Even NV6800 doesn't look as much of flow killer as some of these chipset cooler design I see, so... of course I will always hold my judgement till somoene actually TRIES it out and shows comparison result in same system.

[Radical_53]

Well, I've always measured my GPU temps on the back of the card for as long as I had a Radeon. Now, with the 6800 there's too much stuff on the back. So now, I put my sensors aside the die, close enough that the sensor touches the core.
One thing I'm curious about: How come that with less flow your water temp increases? I've never seen any impact of flow on water temperatures, only some water blocks show different results with more/less flow, at least as far as I know.
Sounds interesting so far, how's your radiator build?

And yes, I agree with you that a flow killer can't really be judged just by its looks. I had two different blocks on my 6800, one which looked really restrictive, the other one looked like good flow. Measuring both of them, flow was basically the same (difference was less than I could measure correctly).

[jinu117]

Well, I've always measured my GPU temps on the back of the card for as long as I had a Radeon. Now, with the 6800 there's too much stuff on the back. So now, I put my sensors aside the die, close enough that the sensor touches the core.
One thing I'm curious about: How come that with less flow your water temp increases? I've never seen any impact of flow on water temperatures, only some water blocks show different results with more/less flow, at least as far as I know.
Sounds interesting so far, how's your radiator build?

And yes, I agree with you that a flow killer can't really be judged just by its looks. I had two different blocks on my 6800, one which looked really restrictive, the other one looked like good flow. Measuring both of them, flow was basically the same (difference was less than I could measure correctly).

How did u get into core? Got some of the shim off? I have Weapon heatercore (the 2 120mm fan one). Actual restrictiveness only comes from GPU and CPU block really per say considering my pumping power. (dual mcp600) Now since both GPU and CPU are heavily restrictive, what I would guess is happening is waterflow drops more than by half (certain threshold) by activating only single pump thus rendering radiator less effective. Thus water temp going up. (not as much as block though) Unfortunately when people do review on waterblock for flow vs performance, the peformance difference have to include efficiency of radiator based on flow. Reason why I like pro-cooling review as they do delta of water temp vs block temp instead of ambient vs block temp. Depending on how restrictive the setup is, there is big difference on temperature with pump. I once tried external watercooling loop with eheim1250 and that actually increased temperature with same comoponents by 2c due to extra 4ft of tubing. Now I understand some components are better designed in mind with lower flow than others but still less flow = less performance. I would love to see someone review this though... just for hell of it

[stu_allen]

I don't know what GPU clocks you are getting, but an FX-55 @ 2760 on water and 46C under load is plenty reason to want more flow. You could be doing better with the stock air cooling. When stock air is better than your water cooling, you can't say you don't need the flow. In your case, flow is not the only thing it seems you need. I'd imagine you are in it for looks/noise, not performance?

Noise mostly. Dont care about looks - no windows in my case
46 is cpu temp using the mobo diode, but i'm rebuilding at the moment using the dfi board instead of the asus, so i'll thermal expoxy a temp sensor onto the copper base of the waterblock if i can.

Clocks were very much limited by the asus motherboard not being able to supply enough vcore, not to mention reluctance to clock tccd over 240.

In terms of chipset waterblocks, both of these koolance blocks fit just fine with 2 6800GTs and aquagrafx waterblocks.
GPU-180-H06 (Vid)

http://www.koolance.com/shop/product...products_id=87

CHC-A05 (MB)

http://www.koolance.com/shop/product...products_id=86

Ive ended up using the latter, as it has very little flow restriction (the top one has quite a bit more, but a lot more surface area too) and is generally easier to work with for the position of the barbs.

[Radical_53]

@jinu: I have a dual 120mm radiator too, I'll have to check how yours looks Interesting thing. Still, normally only the blocks can improve with more flow, due to their design. Also, as you said, your temperatures went up, just as the water temperature. So your delta stayed the same basically, right?
And yes, I agree with you that the only "correct" testing method can be to measure the delta, not the "overall temperature" itself, as any good block would suffer from this (it can take away more heat, water temp is raised, overall temperature goes up).
Also, I saw the same thing like you did with too long tubing. I once re-worked my whole setup and shortened all tubes, put the whole thing together in a different way (blocks stayed in place). Improvement was roughly 5°, without changing any of my cooling components.
About the measurement: My sensor is only "touching" the side of the GPU itself, it's not in there or anything. I cut out a little piece of the sensors tip and put some thermal grease between the sensor and the GPU to make it a little more accurate.

[saratoga]

The problem with restrictive blocks is two fold.

First, you get lower flow. This obviously hurts performance, but can be compensated for with good design or large pumps (see the NexXxoS).

The second problem is that low flow starts to hurt your radiator performance and you have to compensate with a large radiator (more fans), faster fans or a much larger pump. Either way you start to get noiser just to get the same performance as you could have gotten with a lower restriction block. Obviously if you're not too concerned with performance, this may be acceptable, but then you start to go down the whole slippery slope of water vs. air cooling and cost vs. benefit.

[stu_allen]

The second problem is that low flow starts to hurt your radiator performance and you have to compensate with a large radiator (more fans), faster fans or a much larger pump.

Hmm, not sure i understand that. Surely lower flow *through the rad* is good - you transfer more heat to the fins - and then air.
See http://www.thermal-management-testing.com/radiat13.gif for flow vs heat dissipation.

[Holst]

Flow and rads is not necicarally more = better.
Sometimes more flow through rad = WORSE performance.

Read billa's stuff on overclockers.com

[Radical_53]

@saratoga: More flow for the blocks means more performance (no matter how much exactly "more" is), but more flow through the rad?
Doesn't make sense. I've never seen any rad to have different perfomance with different flow rates. As the other two say, it doesn't make any sense

In my case, I use a Laing DDC pump (called Swiftech DDC here, even if it's made in Germany) and a Black Ice Radiator (couldn't find much about the "weapon heatercore" on the net), performance is really important.
As I use the Nexxxos blocks and a pump with good pressure, I won't need too much flow to get a decent performance.

[jinu117]

Interesting, but the water temp increase I get isn't lying to me :P (very easy for me to test since pumps are serial). Maybe single MCP600 with those 2 restrictive block is not enough to even give decent flow for system to hit sweet spot? In which case, chipset block such as mentioned would be as problematic as G4 or silverprop.

[Radical_53]

Well, who knows the reason? Seams complicated, still I can't believe it's the lower efficiency of the rad because of low flow... really, I've never seen anything like that. Do you have a picture or something of your rad?
And, how do you measure your flow? Normally a second pump would not double the flow, maybe give an 80% increase or so. I'll try that as soon as I get my double-DDC case
I know some people that have rads like nuclear plants (not as big, but works just like there), which do really need all the flow they can get.

[HARDCORECLOCKER]

The one in post #1 fits perfectly - if I get a working board some day will tell the temps here...........

[[XC] leviathan18]

why argue about WC start using phase change no rad pump waer block restriction problem.

and with a good made phase unit you can get -20°c

[Radical_53]

Well, that's just what I said here somewhere. If you want to have those really low temps, you'll have to phase chance anyhow.
So the advantage of watercooling has to be something else For me, it's lower temps than air cooling combined with lower noise than air cooling and phase change. For me, it's an sufficient and efficient way of cooling, but nowhere close to break records

[HARDCORECLOCKER]

Can not confirm the waterblock is a "flowkiller" - got nice temps with it, about 10° C lower than using the chipset stock aircooler..........

You can check it here:

<img src="http://img174.exs.cx/img174/4557/85192kn.jpg" alt="Image Hosted by ImageShack.us" />

[Radical_53]

No, it's definitely not a flow killer. For my system, flow wasn't reduced by it, temps are quite good (never used stock though), ~30-33°C under load.

[TheMeatFrog]

Muhahaha I'm bringing this thread back to life So what kind of options do we have now for chipset waterblocks on the DFI nf4s?