Watercooling an Athlon FX-55 and dual Nvidia 6800 GT's (SLI)
 

I am planning on purchasing a new SLI motherboard, 2 x Nvidia 6800 GT's, and an AMD Athlon 64 FX-55. I think I have plenty of room inside the case (Chenbro SR107) for most cooling setups. This will be my first Watercooling system.

I was thinking of getting one of the following setups:

Swiftech MCW6000-64 CPU Waterblock (3/8" ID)
Swiftech MCW20-B Chipset Waterblock (3/8" ID)
2 x Swiftech MCW50B GPU Waterblocks (3/8" ID)
2 x Swiftech MCW50-NV4045KIT Adapter Kits for NV6800
Swiftech MCP350 WaterPump (3/8" ID)
Black Ice Xtreme II Radiator (3/8" ID)
5 1/4" Bay HDP Reservoir (3/8" ID)
Tygon Tubing (3/8" ID)
32oz of FluidXP+ Coolant

-or-

DangerDen Copper TDX CPU Waterblock (3/8" ID)
DangerDen MAZE4 Chipset Waterblock (3/8" ID)
2 x DangerDen 6800 GPU Waterblocks (3/8" ID)
DangerDen DDC-12V WaterPump (3/8" ID)
Black Ice Xtreme II Radiator (3/8" ID)
5 1/4" Bay HDP Reservoir (3/8" ID)
Tygon Tubing (3/8" ID)
32oz of FluidXP+ Coolant

Which one would be better and why? What else am I missing? Will the MXP350 be powerful enough for either of these setups? Would you recommend other items instead of the ones I chose? Am I crazy to think I could watercool these monster heat generators with the above setups?

My case :

http://img.photobucket.com/albums/v3...-SR107-ATI.jpg

I'm planning on mounting the Radiator right beind my two 120mm intake fans in the center of the case. The case is much bigger than it looks, it has spaces for two 92mm fans up front in the drive bays, 2 x 120mm fans in the middle and 1 x 120mm exhaust fan.

The motherboard I am planning on getting:

http://usa.asus.com/products/mb/sock...a8nsli-d_l.jpg

If all this is possible, what route would you recommend that the water takes through the system?

Will 32oz of FluidXP+ be enough for this system?

I'm not even close to being a watercooling expert but I think you would be better off going to 1/2" tubing, and if you plan on cooling it all in one loop maybe run 2 pumps?

Oh and when you do get it setup, please post pics. :)

Wouldn't using 1/2" ID tubing on a loop where most of the components are 3/8" ID including the Pump and Waterblocks be detrimental to the flow or cause a bottleneck?

I could use two pumps, I'm just not sure how to route all this stuff in my case :(

I value quietness. I currently have 3 x 120mm Globe fans known for their lower noise floor (34dBA), I really don't want anything louder than those if possible so I'm assuming the D4 Pump is out of the picture. I've heard a recording of the D4 in the DDS Post, and if its anything like it is there, its way too loud.

-Josh

If I were you I would get the Swiffy GPU blocks instead of DD blocks (very heavy, will hurt your flow a lot, and you don't really need to water cool your ram, passive heatsinks will do just fine) I am using the swiffy block on my 6800 ultra along with passive heatsinks and I can overclock my GPU to 475/1.18 and my temps are great (38-42 idle and 42-45 load depending on the room temps) oh and most of all you will save enough money by doing so. I also think you should get a second Swiftech MCP350 pump for better performance. one last thing, dual heatercore with your setup will be better than Black Ice Xtreme II Radiator that's if space is not a problem tho.

good luck

I"d go wiht the swifty setup because teh 6000 and mcw have a lower pressure drop (i think).

with that many blocks you will probably want to get the swiftec stuff. The 6000 doesnt take too much flow, however you may still want to look into a bigger/second pump
the danger den (especially tdx) will kill your flow, you will certainly need a bigger pump for that.

And I would definatly nix the fluidxp, imo it's a waste. Just do your system right and you won't have leaks. Spend the extra money on cooling/rad. Dissapating that much heat will need a pretty big rad/heater core with 2 or more fans on it

I would get a RBX for the CPU and take each outlet to each graphics card and then to seperate inlet on the res. I would run 1/2 tubing to and from the pump and 3/8 from the cpu block to the GPU block and to the res. From there I would use !/2 again. I would use the Maze4 gpu blocks when they get the new brass tops which will work on the Nvidia cards.

i don't know if that pump alone could handle that system what if you make to seperate loops?

get the mcp600 or 650, you could really use the extra flow.

swifty mcw 6000 x 1
dd maze4 GPU x 2
swifty mcx-159 x1
mcp 350 x 2

for rads, you mentioned liking low noise which means low flow low pressure fans and Cathar has done some tests showing that unless you are using some strong fans then there is no point to using a rad more than 1" thick so I would say get

BIPII x 1 (inside front)
BIP x1 (rear)
Radbox for the BIP

BTW, what kind of case is that and can you post some more pics of it?

my personal recomendation would be:
MCP-600 (now known as aquaxtreme 50-z i think)
MCW-6002
Silverprop cyclone fusion HL (or the SL version of there isnt enough vertical space for teh HL)x2
and a heatercore or a pair of BIX rads.

the reason for the mcp-600 is it is quieter than teh mcp 650 and has slightly higher head pressure than teh mcp-650 too, i must admit it looks better than the mcp-650 too.

run the whole thing in 1/2" tubing as it will all take 1/2"

also your case has ALOT of space compared to mine, im shure you can get all that to fit with some minor modding.

first off let me say nice case selection :D For my last xeon setup I wanted that Chenbro but ended up getting a cheap Intel 5U case for 40$ delivered, same size but not compartmentalized (sp? :shrug: ) like yours is, quite nicely too.

I hacked at my 5U case soooo many times trying to figure out the best way to mount a double rad. I have to shoot down psychofunk's idea because I sort of tried that already. I had a double exhausting up top and tried a BIX infront. Either way you do it the air going over one radiator is also going over the other. Really did nothing for my temps and just added TONS of clutter.
Putting the rad infront of those two 120s would be easiest for your case, but if you have time and like hacking stuff up Id go another way :D
Take out those two HDs you have and put them under your CDROM, seal off that top part as much as you can so your PSU pulls air over your HDDs only.
Then hack out HDD cages as much as possible so you can just put a double 120 HC right there in the front right by the bezel. Having those two 120s suck through the HC/radiator would be more efficient than blowing through them, more fins would get airflow. That would give you like a 5 inch plenum, or however deep those HD compartments are. Just seal up as much as you can so the only place your 120s can pull air from is through the rad. having that plenum and not having the fans RIGHT on the rad will keep the noise down also. Id cut out those grills over the 120s too, just adding air restriction.

To give you a general idea: http://tinypic.com/wiqvc

As for pump Id definitely say mcp600, maybe even two if you arent strapped for cash. Two mcp350s would be a waste. You arent low on space so not really any reason to go for them...
A MCW6002 because of price and performance, and two maze4 GPU blocks because of low pressure drop, performance, and price. NB block maze4 also, lowest pressure drop but not really priced well =\

Are you really gonna blow the hot air from your rad into your case?

Hmmm...

This would be my first watercooling setup and I'd like to try and not hack up a $200 case, if at all possible. I also need active cooling for my 15K RPM SCSI Drives so I wouldn't be comfortable taking them out of their drive bays. My 120mm Fans are rated at 67.28CFM's @ 34dBA. How does this compare to what is needed for a proper radiator setup? I'm kinda limited on space behind the case, so I'd rather not have anything attached to pr sticking out the back if possible. I'm wondering wether I should have gone with the Coolermaster Stacker case... :(

I was under the impression that the Maze 4 GPU Block didn't fit the Nvidia Geforce 6800's? Is this incorrect?

Is there any data on the MCP600 or a review of the unit? How does its noise level compare to the MCP650?

More pics of the case...

http://www.chenbro.com.tw/workspace/..._frontview.jpghttp://www.chenbro.com.tw/workspace/...7_rearview.jpg

I will take some pics of the case when I get some additional time.

Better that than blowing hot air from his case through his rad.

here's a nice pump comparison, in which the MCP600 is recommended very highly for most typical W/C Systems

suggest exhaust rad to outside case, duct room air to cold side of rad...i.e. keep seperate airflow from rest of case

With two ultras and the 64 that's a lot of heat to get rid of. You are going to need either one monster rad or go with a two stage setup. Even with the monster rad, the second ultra is going to be running real hot as you will likely be dumping 50C water onto it from the first one. So your options are to have two seperate loops or put a second rad in between the video cards.
As for sizes, go with 1/2" all the way (port, pump, tubing, rad, res) you will need all the flow you can get.

suspect you are overestimating the temp differences between different parts of the loop..... iirc an o/clocked CPU only gives a rise of a fraction of a degree at any realistic flow rate.

is there any need for a chipset block on these boards?? - maybe that's one way to simplify the flow.

Not sure what you mean by iirc.
I know from experience that my XP3200 adds about 70 watts, translates to about a 8C rise in temp from room temp to give 31C CPU temp.
My 6800GT is adding about the same with the DD NV68 block (GPU and mem) and the temp goes from 31C to 47C GPU temp.
If I where to run this into another video card then that one would be getting water at a temp of 47C.
This is what I've seen on my setup or am I missing something? The ambient temp reported by video card always matches the temp reported for my CPU.
I realize that the water is constantly flowing and being cooled through the rad. However, the heat still has to add up the more stages you put in and the greater the distance the water has to travel before getting cooled again.

iirc = if i remember correctly.....

at (say) 4 litres/minute (just over 1 US gallon/minute and a low flow rate for a system) once a steady state is reached (i.e. the water temperature has risen to the point at which the rad(s) shed heat as fast as the pump+blocks add it) a 70 watt CPU would only raise the water temp by approx 0.25C i.e the temp after the block would be a quarter of a degree higher than that before it.

{specific heat capacity of water is 4.186 joule/gram °C, 4 litres of water weighs 4kg so 66.667grammes per second flow through the block. therefore since a watt = 1 joule/second it would need about 279 watts to raise the water by 1C}

Ok I'm new to water cooling but let me see if I understand what your saying. Once the temperature has reached the "steady" point that in this case it would take an extra 279 watts to increase 1C. I could see this working in a setup with just one heat source and one rad. However, the affect of multible heat sources in series changes this calculation as the heat is increasing at each source with only one rad to cool. The water coming off the CPU is warmer than what is going in correct? If not why does it matter which direction you run the water through the block? That warmed water is then going into the GPU block. Each stage heat is added will have greater affect. If your theory applied then it also would not matter what order you ran the water Pump, CPG , GPU, RAD, Pump or Pump, GPU, CPU, RAD, Pump. You could even run Pump, RAD, GPU, CPU, PUMP. Yet the first setup is the one most people seem to use because it gives the coolest CPU temps. Atleast this is what I have seen and what others have reported online.

well the same arguement would apply: say in this instance the 2 GPUs produced 30W each, and the pump 10W:
then the difference between the coolest water (directly after the rad) and the warmest (directly before the rad) would only be approx 0.5C... (since the total heat input is 140W)

it is better to pick the order that gets you the easiest job of routing the tubing...

It seems to me you have some major faults in your logic:

In your calculation you use 4 liters as the refference. Which is just over 1 US gallon. I don't know what kind of system you've got but mine does not hold anywhere near that kind of water. Even with the res I'm only using 3 1/2 cups of what and then less than 1/2 cup of water wetter. So lets just say 4 cups total. That's just under 1 liter. So though it might be moving at 4 liters per minute, it is still only 1 liter of water. Also you assume that the system can reach a point were the heat is removed as fast as it is added. This may never occur, depending on effeciency of the rad and the alrflow of the fan.
Lastly and again, you are looking at the system as a whole not an individual point in time. The heat transfer is not happening equally throughout the whole loop. The water is heated and cooled at set points in space and time. Thus the heat is not instantly teleported to the rad to be shed. In the multi-heat configuration you have a series of heat sources that cascade heat into the loop and only one point at which it is shed.
Equations or great on paper but they don't always work in the practical world.
Take my system for example, worse case I have about 148 watts being generated. By your calculations my sysetm does not have enough wattage to increase the water temp by 1C and yet it does and yet it can and does increase more than that.
There are many other factors that have to be taken into account. The ability of the block to transfer heat, rad/fan effeciency, water flow or slow downs, tubing size and length etc, etc.

the actual amount of liquid in the system will only affect the time it takes to reach equilibrium, not the equilibrium temp. the heat sources add heat at (virtually) the same rate irrespective of water temperature - but the rad(s) ability to shed heat is proportional to the temperature differential between the water and the air
http://thermal-management-testing.com/ThermoChill.htm

so the water temperature will rise until it reaches the point at which the rad sheds heat as fast as the pump/blocks add it.... the more liquid in the system the longer it will take, but we're talking minutes unless you have an enormous res.

once it reaches equilibrium it doesn't matter whether 1 litre circulates 4 times a minute or 4 litres once a minute - a smaller volume heated and cooled more often is just as good at moving heat from one place to another.