Watercooling setup Opinions and Reccommendations Appreciated

[noother]

First I would like to say I respect all comments and appreciate any advice people have to give. Also a little disclaimer because I know people will start criticising some of the things I have listed here:

I know I do not need to Watercool my RAM, the Northbridge, or the Hard Drives. That loop is really just for the 'bling' factor and to help keep the overall case and board cool. Besides when OCZ's high end memory costs half as much as the same Corsair Dominator RAM, gotta go with it (and I am a huge, huge corsair fan, I have their RAM in all my other rigs). And since it has the XLC system, why not use it in a separate loop for fun?

Now onto the setup:

eVGA 680i mobo with danger den waterblock on NB
OCZ Flex 9200 XLC RAM
D-Tek Fuzion CPU waterblock on E6600
D-Tek Fuzion GFX waterblock w/ heatsink solution on (2) eVGA Superclocked 8800 GTX's
(2) WD 150 Raptor X's in Raid 1 w/ Danger Den Aqua Drive waterblock
Laing DDC w/ Petra'sTech DDCT-01s Top Combo
Black Ice GTX 480 w/ SilenX 120x120x38mm - 18dBA - 90CFM iXtrema Pro Fans
D-Tek Pro 120 Rad

Loop1:
Res->Pump->BIGTX480->CPU->GPU1->GPU2
Loop2:
Res->Pump->BIGTX120->NB->RAM->HD

I have heard people really like using the swifty waterblock for the GPU. I am seriously considering that with the heatsink kit. I recently had some people saying they really liked their D-Tek GPU waterblock and full heatsink cover combo. That in turn made me also look at their single 120 Rad, since that loop isn't really heat intensive imho.

What I would really appreciate is some comments on better efficiency waterblocks, loop setups, tips n tricks, etc.

[Ad1tya]

PA120.3 would be a better choice.

PA120.3 is better than the GTX480. Youd need noisy fast high CFM fans with the GTX480 to perform on par with the 120.3
GTX120 seems cool for the NB/RAM/HDD.

Stay away from the Fuzion GPU block, its a flow killer. Get the MCW60B. Much better. Specially with so much in your loop.

Though, Id advice keep your Graphics cards in a Seperate loop, and your CPU + NB + RAM + HDD in a seperate loop.
PA120.3 for CPU, NB, RAM, HDD.

PA120.2/120.3 for 2 GPU's.

[noother]

PA120.3 would be a better choice.

PA120.3 is better than the GTX480. Youd need noisy fast high CFM fans with the GTX480 to perform on par with the 120.3
GTX120 seems cool for the NB/RAM/HDD.

Stay away from the Fuzion GPU block, its a flow killer. Get the MCW60B. Much better. Specially with so much in your loop.

Though, Id advice keep your Graphics cards in a Seperate loop, and your CPU + NB + RAM + HDD in a seperate loop.
PA120.3 for CPU, NB, RAM, HDD.

PA120.2/120.3 for 2 GPU's.

That is interesting, I have actually heard from quite a few people that the 480GTX is better than the PA120.3. Also with the fans I have speched for the system, you aren't talking about that much noise. I have also seen a decent independent review comparing the PA120.3s to the Black Ice Rads, and the Black Ice Rads outperformed the PA120s. Honestly they were all fairly close, the PAs are very good Rads, I just happen to really like the revised version of the Black Ice Rads. But I have also seen a number of people stick with the PAs. I think it also just has a bit to do with product loyalty for some.

If I put the CPU on the same loop as the RAM and HDD, isn't that a lot of restriction going from 1/2" to 1/4"? I thought I would get better results keeping it on the loop with the GPUs with 1/2" and if needed just adding a single radiator inbetween the GPUs. I have heard a few opinions about this, one is that using the single 480 on the loop with the CPU and GPUs is more than enough. Others have said they put a Rad between the CPU and GPUs. And a few experienced LCers have said that its more efficient to put a single Rad between the 2 GPUs.

[septim]

yes too much restriction going big to small tube.
rad between blocks just adds more restriction to your loops.

the 2 loop solution if viable in your case would prove to be much more effective.
gtx480 and pa120.2 if you will. theres one thread here that fitted both in a tj07 bottom.

[Marci]

I have also seen a decent independent review comparing the PA120.3s to the Black Ice Rads, and the Black Ice Rads outperformed the PA120s.

Link?? S'a new one on me...

[DepTi]

Nobody has mentioned it yet but you're wasting your money going with those SilenX fans. They aren't really gonna give you the noise level they say they are. Get some Yate Loon SMs and you'll be better off and save a lot of money.

[Hard H20]

That is interesting, I have actually heard from quite a few people that the 480GTX is better than the PA120.3. Also with the fans I have speched for the system, you aren't talking about that much noise. I have also seen a decent independent review comparing the PA120.3s to the Black Ice Rads, and the Black Ice Rads outperformed the PA120s. Honestly they were all fairly close, the PAs are very good Rads, I just happen to really like the revised version of the Black Ice Rads. But I have also seen a number of people stick with the PAs. I think it also just has a bit to do with product loyalty for some.

noother,

Product loyalty has nothing to do with it, there is no debate, the PA120.3 is king of the hill right now. I know there are those who can't use the 120.3 due to space limitations when doing an internal build. I would recommend that you make the 120.3 fit you will achieve better cooling results all the way around versus using 480GTX. You have a link on the review you mention????

Also Petra did a down and dirty quick run with the FuZion GTX gpu block. The pressure drop was pretty bad. You might want to reconsider using those blocks in your loop if you're looking for performance over bling.

[noother]

yes too much restriction going big to small tube.
rad between blocks just adds more restriction to your loops.

the 2 loop solution if viable in your case would prove to be much more effective.
gtx480 and pa120.2 if you will. theres one thread here that fitted both in a tj07 bottom.

I am not 'as' concerned about space limitations. I am more concerned about overall cost and performance. As for space, design, layout, etc. I can mod or build my own case to handle those. Right now I am looking at purchasing a MountainMods UFO case which should hold everything. With a 120mm Rad between the GPUs, I do not believe restriction will be much of an issue. I know several people with a Rad in their loop using just the D5 pump and are having now flow issues. I can't image with the petra mod pump I would see any problem with running 2 Rads the CPU block and 2 GPU blocks on a 1/2" loop. Do you really think that is a problem, or just maybe overkill?

Nobody has mentioned it yet but you're wasting your money going with those SilenX fans. They aren't really gonna give you the noise level they say they are. Get some Yate Loon SMs and you'll be better off and save a lot of money.

I will look into those. I already have some fans that will work, just looking for a better combo of 90-120 CFM with low noise. They were the last thing on my list to research honestly, they are usually the cheapest and easiest things to replace.

noother,

You have a link on the review you mention????

Also Petra did a down and dirty quick run with the FuZion GTX gpu block. The pressure drop was pretty bad. You might want to reconsider using those blocks in your loop if you're looking for performance over bling.

Definitely I will see if I can find that link, it should be saved on my home PC (at work atm).

I have read a number of Petra's reviews and that was one of the reasons I originally decided against the full waterblocks (DangerDen and Innovatek), and was going to go with the swifty block. I have seen many good comments about the swifty block and was only going to use the D-Tek since I was already using the D-Tek cpu block.

[Grinch]

noother...

to xtremesystems

[phelan1777]

WELCOME TO XS!

Home of the truly Xtremely Insane

<<<<Hands NOOTHER a straight jacket, "please check your sanity @ the door, don't mind the ones in the White coats, they think they are Dr.s when they are the ones truly insane


I am running four blocks on a PA120.3 G5 (Restrictive), dual MCW60s and a MCW30 on my EVGA NB, I get awsome temps for having a QCore and it being OCED. Not to mention the 8800GTS cores. That give off quite a bit of heat as well.

I use three SAN ACE/Denkis pulling, so they are a bit over kill compared to the Yate Loons, but if you have the room, are worried about temp load and have a decent ambient temp, then you could give these a shot.

[noother]

Thank you for the warm welcome. I had to wait a week or so to get the final approval before I could create any threads. I have been doing high-end passive heatsink and air cooling and overclocking for sometime, just now switching over to liquid cooling although I have been reading up on it for awhile as well as the TEC, phase change, etc.

Some links dealing with this issue, read the comments by Ira-K. Ira-K is the one that had been helping me the most with the initial setups and decisions. I have now just been looking for ways to tweak it as much as possible short of building an evaporator (which I have also been considering) or going to phase change (which I think is overkill and too pricey).

Link about PA vs GTX

I am still looking for the actual testing link that showed some of these results. As mentioned by the PA rep, its the higher CFM fans that help the GTX pull ahead. Since I plan to use ~100 CFM fans, it gives the GTX the edge with an okay footprint. I am however not opposed to using lower cfm fans and the PA 120.3 if it is less noisy and costs less.

Also as another addendum, the final product needs to look good, be fairly portable (enough just to take it to a LAN party every other month), and perform well.

[Ad1tya]

Well, whatever Rad you get, Please dont get a Fuzion GPU Block.
That will kill your flow no matter what.

[noother]

Okay I will promise, I won't use D-Tek for the GPU block

I will just use the MCW60 with their heatsink kit.

[phelan1777]

Okay I will promise, I won't use D-Tek for the GPU block

I will just use the MCW60 with their heatsink kit.

Some advice for mounting the MC14s if you use them, use a dab of super glue on the corners, it holds well if done right. (my .02)

[Snyxxx]

Just some words from experience. I had a 240GTX after I had a BIP2. The GTX was worse than the BIP2 for mainly two reasons:

Higher fin density - They added a whole lot more fins (thinner though) for better cooling, sounds great, but the reality was it collected dust a lot quicker. Do not underestimate the power of dust from increasing your temps.

More flow restriction - I cleaned both with vinegar (back in the day when it was cool) and my informal fill with funnel test clearly showed the GTX had a hell of a lot more restriction than the BIP2. The vinegar slowly went down the funnel with the GTX versus rushing down the funnel as quick as I could pour it with the BIP2. The GTX claim of narrow flow channels killed the flow.

Then I went to the PA120.2. I have to agree with everyone that it is the best. The fin density is low and it does not collect dust like either the BIP2 or GTX240. I also noticed the flow was even better than the BIP2 in regards to the funnel test.

I like my rad inside, so I cannot use a 120.3. You cannot go wrong with the PA series. Listen to the voice of experience.

[ranker]

You'll be hard pressed to find room for GTX480 if you do intend to go with a MM case, unless its a custom case. It pretty much means you'll have to end up with a PA120.3 for your main loop and a whatever you choose for your secondary loop. You never mentioned what type of CPU you have. It'd be helpful to know.

I'd suggest a D-Tek Fuzion of C2D or a GTX if quadcore for your CPU block. Since you're putting your CPU/GPU blocks together, I'd have to recommend a MCW60 as well; the EK FC's and D-Tek GFX's are a bit restrictive when thrown in with other blocks.

A PA120.3 should suffice to cool this loop as can dissipate about 600-650W of heat. 400W will be used up by your SLI GPU's leaving you with about 200-250W for your CPU. That should be plenty of headroom for a healthy overclock.

A MCR220 in the back of the UFO should be able to cool everything you have. Heck, I think a 120 rad would work. Just make sure you use a DDC-2 pump in each loop as you have multiple blocks.

[Marci]

Since I plan to use ~100 CFM fans, it gives the GTX the edge with an okay footprint. I am however not opposed to using lower cfm fans and the PA 120.3 if it is less noisy and costs less.

1) You aren't using 100cfm fans, you're using 90cfm fans.
2) Those fans only say 90cfm on the box. They don't shift 90cfm in reality when mounted on a radiator with backpressure imposed by both the radiator and the chassis on the other side of the radiator.
3) The more cfm, the more noise. Lower CFM fans WILL be less noisy, and when it comes to low CFM, the PA wins every time.
4) 3 fans costs less than 4 fans.

I have also seen a decent independent review comparing the PA120.3s to the Black Ice Rads, and the Black Ice Rads outperformed the PA120s

Lol... That is in no way a "decent independant review". That is someone linking to me making a statement on my own forums...

Testing has shown that with 70cfm fans moving at 1200rpm at less, the PA120.3 with 3 of such will match or beat the GTX480 with 4 of such. Move above 70cfm fans and the GTX will begin to pull ahead. Move below 70cfm, and the PA's lead grows.

That is also a PA120.3 singular vs a GTX480 singular... not the entire line of rads which is what was implied by the use of plurals...

the Black Ice Rads outperformed the PA120s

That's why folks here are jumping to defend the PA120.3. The way you've phrased what you said is a bit too unprecise when referenced against your source... which is me anyway...

On the link supplied (overclock.net) folks are asking if one rad with 50cfm fans would beat another with 100cfm fans - that's pointless. Compare rads with the SAME CFM fans...

PA120.3 vs GTX480 with 50cfm fans, PA120.3 wins. With 70cfm fans, PA120.3 wins. Soon as you move to 80CFM fans on both rads, the GTX480 pulls into the lead ever so slightly (less than a degree at 80cfm, and at 100cfm GTX probably wins by 2 or 3 degrees) but you also have the increased space it takes up for those 2 degrees, so there are tradeoffs. But, BOTH rads have the same fans on, and thus produce the same noise. OBVIOUSLY any rad of same footprint or larger, with 100cfm fans, WILL outperform any other rad with 50cfm fans. It is the amount of airflow that determines how much heat the radiator can remove (assuming flowrates are constant)...

[hecktic]

thank you marci for that excellent post

[noother]

1) You aren't using 100cfm fans, you're using 90cfm fans.
2) Those fans only say 90cfm on the box. They don't shift 90cfm in reality when mounted on a radiator with backpressure imposed by both the radiator and the chassis on the other side of the radiator.
3) The more cfm, the more noise. Lower CFM fans WILL be less noisy, and when it comes to low CFM, the PA wins every time.
4) 3 fans costs less than 4 fans.


Lol... That is in no way a "decent independant review". That is someone linking to me making a statement on my own forums...

1) I have 100+CFM fans currently, I was going to buy the silenx fans in addition.
2) This is true, I need fans with more pull, its hard to find the right ones. In air cooling, I have worried more just about total airflow, now I need to worry more about pull.
3) True, but some companies work more on noise reduction and get higher CFM at lower noises than other companies because of it. But the same fan with the same setup will be less noisy at lower CFM.
4) No doubt.

I said I had read an independant review, and then I linked that statement you made saying I was still looking for the independant review. Please don't make passive attacks on what I say without full disclosure.

As I have said, I am merely looking for the more efficient setup for the best price ratio I can get and a good bling factor.

[noother]

You'll be hard pressed to find room for GTX480 if you do intend to go with a MM case, unless its a custom case. It pretty much means you'll have to end up with a PA120.3 for your main loop and a whatever you choose for your secondary loop. You never mentioned what type of CPU you have. It'd be helpful to know.
...
Just make sure you use a DDC-2 pump in each loop as you have multiple blocks.

For the case, I can always mod the case myself to get things to fit, or externally mount the GTX480. But yes this has also been a concern of mine and I have been going back and forth about what to do about the Rads. They are essentially one of the most important components in the setup.

From my original post I mentioned both the CPU and the Pump:

D-Tek Fuzion CPU waterblock on E6600

Laing DDC w/ Petra'sTech DDCT-01s Top Combo

The E6600 is an Intel Core 2 Duo at 2.4Ghz. That particular pump at Petra's shop includes a DDC-2 18W.

I also conceded that I would use the MCW60 with the heatsink kit instead of the D-Tek fuzion GPU block. The MCW60 was my original choice when speching everything out.

My concerns are mainly when I try to get my CPU up to 4Ghz and my GPUs both up to 700Mhz if the Single Rad (whether it is the PA120.3 or the BIGTX 480) will be able to dissipate that amount of heat, or whether I should put a second Rad in that loop and where.

[Marci]

Stick CPU (inc overclock / corevoltage), GPUs and pumps into the ExtremePSUCalculator over here: http://www.extreme.outervision.com/p...ulatorlite.jsp - you'll have to have a ratch to find out thermal output of the GPUs once overclocked, the above only gives stock output, then add the difference on top.

Cross reference the result against Heat Dissipation Graph for the 120.3 over here: http://www.thermochill.com/PATesting/

That'll tell you whether a single PA120.3 will be able to handle the heatload (ballpark) and with what airflow. BlackIce publish no such graphs / data for any of their rads... just a btu/hour figure at an unknown airflow and liquid flowrate. No way to anticipate Thermal Dissipation for the GTX480 basically unless you can find an accurate independant test... but so far, no-one that I'd consider "accurate" has doneso that I'm aware of (BillA or Cooling-Masters). Simple comparisons have been done, but nothing that establishes the various limits (Max Heat Diss or c/w) of the rad at given air and liquid flowrates.

Personally, I'll guess regardless of rad, neither the PA120.3 nor the GTX480 will be able to doso on their own in a low-noise configuration.

[dnottis]

Some advice for mounting the MC14s if you use them, use a dab of super glue on the corners, it holds well if done right. (my .02)

Here is another trick I've been using for mouting ram sinks. Put a dab of arctic ceramique on the GPUs memory, wipe it off. It will actually change the surface from ultra smooth (where nothing sticks) to slightly 'tacky' where the ram sinks stick without a problem. Unlike the superglue you can still remove them safely.

[noother]

Stick CPU (inc overclock / corevoltage), GPUs and pumps into the ExtremePSUCalculator over here: http://www.extreme.outervision.com/p...ulatorlite.jsp - you'll have to have a ratch to find out thermal output of the GPUs once overclocked, the above only gives stock output, then add the difference on top.

Cross reference the result against Heat Dissipation Graph for the 120.3 over here: http://www.thermochill.com/PATesting/

Great, I will put in my figures and see what I get.

BlackIce publish no such graphs / data for any of their rads... just a btu/hour figure at an unknown airflow and liquid flowrate. No way to anticipate Thermal Dissipation for the GTX480 basically unless you can find an accurate independant test... but so far, no-one that I'd consider "accurate" has doneso that I'm aware of (BillA or Cooling-Masters). Simple comparisons have been done, but nothing that establishes the various limits (Max Heat Diss or c/w) of the rad at given air and liquid flowrates.

I agree there is a real lack of real comparisons and true accurate tests for a number of these parts or companies. I guess it just isn't 'mainstream' enough to get more reviews. I have friends here locally that do scientific testing for a living, but they don't have the money to buy all of this equipment or the recognition to get companies to send out units for testing.

Personally, I'll guess regardless of rad, neither the PA120.3 nor the GTX480 will be able to doso on their own in a low-noise configuration.

You mean you do not believe that either will be able to Handle both the CPU and the 2 GPUs overclocked on their own without higher CFM/pulling fans? So then would you recommend a second Rad in the loop, maybe a 120 with less restriction to help out?

[phelan1777]

Here is another trick I've been using for mouting ram sinks. Put a dab of arctic ceramique on the GPUs memory, wipe it off. It will actually change the surface from ultra smooth (where nothing sticks) to slightly 'tacky' where the ram sinks stick without a problem. Unlike the superglue you can still remove them safely.

I am using ASC, it doesn't hold well @ all.

[septim]

hmm phelan forgot to mention that he's running an rd30 pump with those restrictive blocks of his. best comparison or close to it would be to run 2 ddc-2 with aftermarket tops in serial...