Watercooling Config

uZi

uZi

uZi

Associate
Joined
1 Aug 2003
Posts
314
Hi, planning on upgrading to a opteron 170 setup, using a DFI nf4 expert, and also a 1900xt, probs be running g skill ZX ram too

First of all, will my Maze4 fit the 1900? - its around a year old.. maybe a bit less

(im thinking yes, as ive been told the 1900 has the same fittings as a 7800 - which im sure a maze4 fits)

Secondly: is there any point in cooling the northbridge? I'm gonna take the opteron as far as possible, ie 250+ HTT.. will the NB take this with just a stock cooler?

Finally, my current waterpump has developed a crack, so need a replacement, any suggestions? (the current one is a eheim 1250 - but its been pretty crappy.. so probs wont go with one again)

Btw, my current plan for the Watercooling setup will be:

Res -> pump -> cpu -> gpu -> rad -> res

Thanks in advance!
 
Needs to be the maze 4 that fits 6800/7800's

the older revs dont fit. Need to check to see if you have the 6800 mounting kit with it.

Northbridge doesnt have anything to do with the HTT so you dont need to cool it no. although the fan may be a little noisy.

Tom
 
Can't answer about the Maze 4 but wouldn't Dangerden have a compatibility chart?

No point in water-cooling the Northbridge unless you need loud active cooling with an already aftermarket cooler. i.e. if you have the stock cooler and it's not good enough get a better cooler like the Zalman.

I'm surprised to hear you're unhappy with the 1250. Why? Damn good pump with excellent reliability although a tad large. If you can stretch to a Laing D5 (Swiftech MCP655/Dangerden DD12 D5) you will have probably the best pump easily available (the Iwakis are generally regarded as the best but not easy to acquire.)
 
12v pumps are apparently better according to Marci, he has stopped supplying Eheims too. The Maze 4 gpu will definately fit the x1900 with the nv (7800) mounting kit, as opposed to the ati kit. I have found it for sale separately at close to £10 which imo is extortionate for a couple of threads and a backplate. Still cheaper than the latest blocks for x1900/nv7800.
 
Cheers for replies. Will do some checking about the Maze4 guess the 10 quid is gonna make more sense than forking out for a brand new one :)

Nice to know the NB isn't connected with the HTT, i've been running a P4 for the past few years, no idea about clocking AMDs recently.. gonna find out soon i guess :D

@Mike: The eheim has leaked a few times, seemingly for no reason (latest one is a crack at the back, which caused a leak which was about 1cm from taking out my 300gig drive). Also i find it can be quite noisy, even with a lot of bubble wrap under it!! - spose i should get one of those anti-vibration kits Lastly.. it doesnt really fit inside my case very easily, have had it on it's side in my 2 spare 5 1/2" bays, which explains most the vibration i guess. Apart from that i have had it running pretty much non-stop for almost 3 years, so the reliability is good ;)

Will have a look for the Laing D5

Thanks again :)
 
the maze 4 needs to have the correct holes for the mounting kit though.
its not just a case of buying the kit, you still need a correct rev maze 4.

The on die mem controller means that the load is now on the cpu and not the northbridge. Big gains in performance, but generally this will be the weakest part of the cpu now .

Tom
 
welshtom said:
the maze 4 needs to have the correct holes for the mounting kit though.
its not just a case of buying the kit, you still need a correct rev maze 4.

Tom
Click to expand...
Ive heard this too. My maze 4 gpu has 4 holes spaced out in a square, and 2 diagonal holes in the base and another 2 diagonal holes from the anodised aluminium top. If u've got these 8 holes too I reckon u got a later revision.
 
megatron said:
Ive heard this too. My maze 4 gpu has 4 holes spaced out in a square, and 2 diagonal holes in the base and another 2 diagonal holes from the anodised aluminium top. If u've got these 8 holes too I reckon u got a later revision.
Click to expand...


yea that is correct
 
Zefan said:
I dunno if someone has already said this, but it should be Pump>Rad>CPU>GPU>Res>Pump.
Click to expand...

This is not correct!

Please do not repeat this watercooling myth as if it were fact. It isn't and has been demonstrated to be a simplistic statement taking very little else into account. It is far more important to have s mooth loop with as few bends as possible rather than some dogmatic sequence.

Please check in the now archived Water-cooling sticky (when did that happen?!?) for Cathar's excellent debunking of this and see the Overclockers Watercooling Myths Exposed article.

We really must try to raise the "net clue" in this place!
 
Think i have the later revision.. for starters its got the DD logo etched into it, instead of the older design (which i remember from a year or 2 ago).. and im almost certain its got the square of holes for ati and nvidia cards, although i cant check right now as its attached to my gpu ;)

Zefen: It really doesnt make a huge difference, as far as im concerned, you want the coolest water hitting the CPU first, then the GPU, then the NB if its there.. apart from that its really not all that important. Although obviously the water should be coolest when leaving the rad.. personally ive got the res just afterwards, to allow the water another chance to cool.

All of this is in terms of a few points of a degree though, cos the flow is moving too fast ;)

So, basically what mike said ;) - want a nice free flow with no kinks and few bends. A big ass rad never hurt either ;)
 
MikeTimbers said:
This is not correct!
We really must try to raise the "net clue" in this place!
Click to expand...

Water Cooling Myth Article said:
The water temps will rise 1ºC for every 33 watts in the water. So, if your pump dumps 33 watts into the water, water temps will rise 1ºC

Assuming the pump dumps about 50 watts of heat into the water and the flow rate is 1 gallon/minute (gpm - very reasonable assumptions)there is a 0.19ºC difference in water temperature between the inlet and outlet of the pump.
Click to expand...

:rolleyes: Do I sniff a contradiction :\

Although yes I do appreciate the fact that kinking and length are most important, I believe that not only the pump and blocks will be heating the water in this instance. Firstly as water is a pretty good conductor, simply running through a hot case at load will heat it up a treat, especially if the tubing is anywhere near the RAM. Secondly as he is running with a reservoir, this will add more length to the loop thus more time for it to heat up. If you take into account these other factors then it's clear it's in his best interests (Secondary to kinking etc of course) to dissapate heat before it reaches the blocks.
 
Now we have context!

In your context, you are right but you originally stated an opinion with no context as if it must always be true.

As for the article, note I also said check out Cathar's post quoted in the sticky:

Cathar said:
The ordering of the radiator and the waterblock is almost totally irrelevant as the thermal capacity of the water means that it simply does not rise and fall in temperature by more than a few tenths of a degree C as it moves around the loop.

Having said that, the optimal way to set it up, if one is so inclined to be picky about it, is

reservoir->unrestricted opening into pump->radiator->waterblock->reservoir

or for closed loop systems, omit the reservoir.

However, if plumbing routing constraints imposed by spatial limitations means that it's easier to do pump->block->radiator, then do that instead. The only difference between the two is the pump heat gets added to the water before the water enters the CPU. The rise in water temperature via pump heat added per pass through the pump is almost always going to be less than 0.1C for almost every water-cooling setup I can think of.
<snip>
Water roughly has a thermal capacity of 4200J/kgC. Water roughly has a density of 1.0. This means that at 1LPM, per second the water can absorb 70W of heat and rise by just 1C. For low-flow systems with Eheim 1046's pushing a system-wide flow rate of 1LPM, the pump heat is around 3W or so, or around a 0.05C difference as the water flows through the pump. For those with the very strong Iwaki MD30-RZ's, pushing 10LPM through a system and the pump is dumping ~50W of heat into the water, we're still talking about a 0.07C difference.

In short - the ordering of the components is trivially unimportant.
Click to expand...

Note the last line although I concede he says pump>rad for a "perfect" setup if it doesn't cause other problems.

I have never heard on any forum anyone arguing that the case heat warms up the water. To heat something the source must be warmer than the target. A case hot enough to add heat to the water would have to be seriously warm. Where could that heat be coming from? RAM? I doubt it could possibly be sufficient to cause heating of water inside tubing. Should the res be insulated against case heat too?
 
Well said Mike!

As a self appointed watercooling expert maybe you should make a web page you could link on your sig (and I'm sure I would too) to debunk some of the myths.. the overclockers.com article on this is brilliant HERE

My personal favourites that are often pimped round here..

overclockers.com article said:
You shouldn't even ask what someone else's temps are. This is one of the biggest problems I see in forums. YOUR temps are not even accurate relative to your own system, much less someone else's.

Let me elaborate: Just because your board reads 41ºC, it does not mean that if the temps drop 5ºC in real life the board will read 36ºC. Motherboard temperature sensors are not "absolutely accurate", nor are they "relatively" accurate.
Click to expand...
And
overclockers.com article said:
The water in a watercooling cannot get any lower than the temperature of the air around the watercooling system flowing through the radiator. In order to make the water colder than the surrounding air, energy must be applied either in the form of a compressor system or a thermoelectric device, both of which will cool the water, sometimes below ambient. The temperature point at which water condenses is a function of both temperature and humidity.
Click to expand...

So many people are happy to say they've got cpu temps of 20 something when unless they're outside its almost impossible.. Rathr than bigging myself up I'd be worried what they ACTUALLY were!
 
matt100 said:
As a self appointed watercooling expert maybe you should make a web page you could link on your sig (and I'm sure I would too) to debunk some of the myths.. the overclockers.com article on this is brilliant HERE
Click to expand...

Hmm, not sure about the expert part. Most of what I think I know I got from places like procooling where the real experts are. There's plenty of material around for those who are prepared to look for it. The sticky on these forums isn't a bad place to start which is why I put it in my signature. If only more people would read it!

As for the overclockers Myths article, zefan makes a good point above. It is actually contradictory so although its motives are good some of the actual detail is wrong. The article was written as a consensus over on procooling and as is always true, anything done by committee results in fuzzy thinking.
 
You must log in or register to reply here.
Back
Top Bottom