Trifire 6970 cooling

[USMCGrunt]

Would a single 120mm rad be good enough to dissipate the heat of 3 6970s or would I need two 120mm radiators?

 

[QuiteSufficient]

You generally need 1 120mm radiator per high-end GPU, if you want to achieve better temperatures than the factory air cooler provides.

 

[USMCGrunt]

You generally need 1 120mm radiator per high-end GPU, if you want to achieve better temperatures than the factory air cooler provides.

Thank you sir...drawing up plans for an external enclosure for 2 120mm rads, 4 fans, pump and res. I've read the mcp665 is a strong pump, any comments on that?

 

[Aus10]

Thank you sir...drawing up plans for an external enclosure for 2 120mm rads, 4 fans, pump and res. I've read the mcp665 is a strong pump, any comments on that?

Um I wouldnt use anything less then a 360 for 3 gpus if not more... and yes the 655 is a good pump

 

[Tsumi]

2 120's won't net you any better temps than stock cooling with the same amount of noise, so it'll be a completely worthless venture.

 

[USMCGrunt]

I'm mostly looking to keep the cards under 100c (which they are hitting now) and not have it sound like I have a jet engine in my computer. I think the stock coolers would work fine if they weren't sandwiched against eachother..

 

[FaRKle0079]

A 240 radiator will sufficiently meet your objectives of <100C and not be ridiculously loud. If you want to get significantly better temperatures than air cooling you'll want to move up to a 360 rad.

 

[Tsumi]

If you're doing an external enclosure, just move it up to a 360 radiator, if not 480. Why skimp out when you have a lot of room to work with? You will probably be able to make it much quieter also, since you're already dropping more than $300 on waterblocks (assuming that you're getting fullcover blocks), why not spend ~$50 more for a much quieter setup?

 

[magoo]

Single 120.3 radiator should do fine.

I have my CPU and a 6990 cooled by a 120.3 and the load temps on the 6990 (OCd) are about 48C.

Look into using a simple RadBox.

 

[SIRTIM3]

I have a mcp655 with a Swiftech MCR320-QP cooling my 3 6970's (6950's fully flashed) and its perfectly fine with low rpms fans on rad. Cards avg. mid 50's celsius 3rd card in loop is always about 8 degrees hotter. My setup is virtually silent, going with a thicker rad then mine can get you even better temps!!!!

 

[MD]

Hi,

Every radiator has heat power dissipation rating. This rating is many cases over rated due fact that for such heat dissipation one would need have to high water flow and to much (to loud) air flow.(Koolance claims 400 watt heat dissipation for 120 rad with 44 millimeter thickness and brass tubing)

I use generally 60 millimeter thick radiators. They are water flow friendly and can dissipate more heat per fan spot (if provided with enough airflow regarding to a fact that with same fin density they need more air pressure). Depending on fan choice and fin density as well as water flow they can dissipate any were from 300 to 450 watt heat (withing comfortable limits and reserve) per one 120 fan spot.

Black Ice GT Xtreme series radiators have very high fin density of 20 FPI (fin per inch) and ask for fans with higher air pressure and mostly higher rpm's. With proper fan they can transfer ton's of heat to air. On other hand they have less parallel tubes so they produce higher pressure drop then some other radiators.

On other hand XSPS RX series Radiators come with 8 FPI and can use even very slow fans to achieve good cooling. Also they are easier on pumps and make less pressure drop in loop.

I tend to use 10~12 FPI density rad's cooled with 1k to 1,4k rpm fans with higher air pressure.


OK than back to your situation. You want to cool 3x HD6970 cards that on stock can produce up to 750 watt heat. OC-ed they can hit 1KW+ heat. For that kind of a heat load you will need minimum 3x120 radiator that is 54~63 millimeter thick. That said water flow should be around 180 L/ hour and fans should push 55+ CFM effectively true radiator. 55+ CFM asks for 3 fans of 30+ CFM and enough air pressure. Silverstone AP120, Noctua NP12 comes in mind. Also 180 L/h does not mean pump of 180 L/h as pump that can push 180 L/h effectively true 3 water blocks, radiator, reservoir and tubing. Depending on tube length and position that means Laing DCC (with good top) or D5 class pump.

Personally I would go with XSPC RX480 radiator with low noise fan's, I would go with EK FC6970 Acetal + Nickel water block (new revision). I would use XSPC Dual bay DDC reservoir and DDC355 pump. I would try to go with parallel water blocks to avoid overheating last card in chain. I would try to position reservoir as high as possible. I use ThermoChill EC6 coolant (non conductive, with all needed additives). I would use 1/2 ID tubing and fittings. I use Tygon silver tubing in my loops. I go mostly with inline temperature sensor as well with fan controller (Kaze Master or something like that) as well with some solution for maintenance (fill port and drain tap).


Good luck


MD

 

[USMCGrunt]

Hi,

Every radiator has heat power dissipation rating. This rating is many cases over rated due fact that for such heat dissipation one would need have to high water flow and to much (to loud) air flow.(Koolance claims 400 watt heat dissipation for 120 rad with 44 millimeter thickness and brass tubing)

I use generally 60 millimeter thick radiators. They are water flow friendly and can dissipate more heat per fan spot (if provided with enough airflow regarding to a fact that with same fin density they need more air pressure). Depending on fan choice and fin density as well as water flow they can dissipate any were from 300 to 450 watt heat (withing comfortable limits and reserve) per one 120 fan spot.

Black Ice GT Xtreme series radiators have very high fin density of 20 FPI (fin per inch) and ask for fans with higher air pressure and mostly higher rpm's. With proper fan they can transfer ton's of heat to air. On other hand they have less parallel tubes so they produce higher pressure drop then some other radiators.

On other hand XSPS RX series Radiators come with 8 FPI and can use even very slow fans to achieve good cooling. Also they are easier on pumps and make less pressure drop in loop.

I tend to use 10~12 FPI density rad's cooled with 1k to 1,4k rpm fans with higher air pressure.


OK than back to your situation. You want to cool 3x HD6970 cards that on stock can produce up to 750 watt heat. OC-ed they can hit 1KW+ heat. For that kind of a heat load you will need minimum 3x120 radiator that is 54~63 millimeter thick. That said water flow should be around 180 L/ hour and fans should push 55+ CFM effectively true radiator. 55+ CFM asks for 3 fans of 30+ CFM and enough air pressure. Silverstone AP120, Noctua NP12 comes in mind. Also 180 L/h does not mean pump of 180 L/h as pump that can push 180 L/h effectively true 3 water blocks, radiator, reservoir and tubing. Depending on tube length and position that means Laing DCC (with good top) or D5 class pump.

Personally I would go with XSPC RX480 radiator with low noise fan's, I would go with EK FC6970 Acetal + Nickel water block (new revision). I would use XSPC Dual bay DDC reservoir and DDC355 pump. I would try to go with parallel water blocks to avoid overheating last card in chain. I would try to position reservoir as high as possible. I use ThermoChill EC6 coolant (non conductive, with all needed additives). I would use 1/2 ID tubing and fittings. I use Tygon silver tubing in my loops. I go mostly with inline temperature sensor as well with fan controller (Kaze Master or something like that) as well with some solution for maintenance (fill port and drain tap).


Good luck


MD

Awesome, lots of good info here thanks. I was looking at the EK blocks as well. Question though, you mention putting the resevoir as high as possible, why do I want to do this and what am I trying to gain height from, the pump, waterblocks, or radiator?

 

[Tsumi]

Awesome, lots of good info here thanks. I was looking at the EK blocks as well. Question though, you mention putting the resevoir as high as possible, why do I want to do this and what am I trying to gain height from, the pump, waterblocks, or radiator?

He has some good information, and some slightly misleading information, most of which are popular watercooling "myths."

The radiator information is all well and good, but that "wattage dissipation rating" does not mention at what fan speeds the radiator will dissipate that much wattage.

Generally, my favorite radiators are the Black Ice GTS ones. Extremely high FPI and very thin, so it is easy to mount in a case while avoiding clearance issues. And it does particularly well with 1500+ RPM fans in pull only, and 1000+ RPM in push/pull, better than the XSPC radiators in that fan speed range while not costing an arm and a leg like the Black Ice GTX radiators.

Placing the reservoir as high as possible is not necessary. The MCP655 and 355 (most common pumps for computer watercooling) both have enough power to push water through any watercooling setup inside a case. They can push water ~7-10 feet high, so the only thing you do need to do is make sure the pump is below the reservoir.

 

[MD]

Hi, USMCGrunt

You are welcome.

I place reservoir up high for few reasons. It is easier to trap air in system. You always go reservoir > pump and if pump is lower you help a bit with a drop (not that much difference but I take what I can get). With pump at low you gain on distance between a pump and next peace in chain but pumps easier pump out than suck in coolant.
Radiators tend to keep some air when they are highest point and horizontal placed in system (though I like top mounted radiators ). Though with some turning, when bleeding air, one can remove most of air out of it. Going res > pump and filling a loop true res makes sure your pump never runs dry. Fill port at top of a PC some tubing and than into top of a reservoir works good as well.

I go mostly res > pump > rad > wb > back to res. Some people go first to wb than to rad to maintain high pressure coming into wb. It works as well and order of wb and rad makes anyways very little difference. I tend to like water to "cool of" before entering wb and heat added by the pump is also negated by radiator (18 watts ain't that much though). In real life water does not change that much it's temp true all loop. Much less than one should think. Water passes true loop so fast (mostly) so all what is important is how much heat are you adding into loop and how much heat can you dissipate with radiators with given loop and given airflow. Next to it one should not obstruct water flow with unneeded kinks and turns in hoses or unneeded L fittings. Yes loop can look much nicer with few L fittings but to much of them specially after each other can restrict water flow to much.

To fast loop can reduce cooling efficiency as well while coolant does not have time to pick up heat or to exchange heat in radiator. this does not happen that often except with simple loops used with powerful serial pumps (few pumps in series)

One can use pump regulators and pumps with regulation on them and find sweet spot. Though most of a times advantage in deg C is payed with lot's of money. Good option is using strong pump > water flow meter with pump regulator/water flow monitor. Or even pump > water flow meter > pump. That way you know whats your flow and you can regulate it Plus negate pressure drop made by ware flow meter.

GPU water blocks do make lot of pressure drop. Going 3 in series asks for strong pump. Going parallel blocks make less resistance overall but pump pressure is divided by 3 so good pump is a must anyway. D5 with regulation is good option. DDC1 plus MCP355 with good top is good as well. You want 4m + of head pressure.

So back to simplicity one of those pumps,res, WB's and 480 rad and 1/2 ID loop will do the job.


MD

 

[Tsumi]

If you do go with 1/2 ID, make sure to get 3/4 OD so that you can route the tubing easier. 3/4 OD is much less likely to kink. If you want to make tighter turns, go with 3/8 ID 5/8 OD. 3/8 has been shown to have minimal (less than .1 C) differences compared to 1/2.

 

[FaRKle0079]

Every radiator has heat power dissipation rating. This rating is many cases over rated due fact that for such heat dissipation one would need have to high water flow and to much (to loud) air flow.(Koolance claims 400 watt heat dissipation for 120 rad with 44 millimeter thickness and brass tubing)

Not true. Plenty of PC radiators don't have any wattage rating due to no standardized testing. Koolance might say 400W, but at what delta is that? Generally WCers use a 10C delta, but I guarantee you that a 120rad isn't going to achieve a 10C delta with at 400W.

You don't need to worry about flow rates with radiators since even the higher restriction ones aren't nearly as restrictive as any sort of block.

Silverstone AP120, Noctua NP12 comes in mind.

AP121 is a terrible rad/heatsink fan for the money.

I use ThermoChill EC6 coolant (non conductive, with all needed additives).

Standalone it might be non-conductive, but as soon as you mix it with a conductor such as water... guess what, it's conductive! Filler fluids are a joke and a scam, nothing is more functional than distilled water + PT nuke or silver.

I was looking at the EK blocks as well.

EK nickel = garbage. There are cases of even their new nickel process flaking. Don't waste your money with a crook like Eddy.

Placing the reservoir as high as possible is not necessary. The MCP655 and 355 (most common pumps for computer watercooling) both have enough power to push water through any watercooling setup inside a case. They can push water ~7-10 feet high, so the only thing you do need to do is make sure the pump is below the reservoir.

It's not essential, but it is more convenient IME when filling up the loop and bleeding.

With pump at low you gain on distance between a pump and next peace in chain but pumps easier pump out than suck in coolant.

This is just flat out bad information. DDC and D5 pumps are NON-SELF PRIMING. That means these pumps DO NOT INTAKE any water on their own and thus need to be gravity fed water. Due to that reason the reservoir should be higher than the pump (or you're asking for a difficult time).

 

[USMCGrunt]

Excellent stuff guys, really do appreciate all the knowledge. I was actually planning on having the system outside of the case so that if/when I move to another case, im not screwing around with mounting inside a case or being limited to ones large enough for the parts to fit into. What I had in mind was something like this:

MCP665 Pump
3/8x1/2 inch tubing (what is the importance in tubing size other than reduced kinking with large sizes)
250ml Resevoir
480 radiator with something like these fans ( http://www.frozencpu.com/products/8..._Fan_-_Low_Speed_D12SL-12D.html?tl=g36c15s562 )


I had already read that some of these pumps are not self priming so I had planned on having a resevoir "higher" than the pump. I also planned on the sequence being pump>blocks>rad>resevoir. My line of thinking is that it wouldnt introduce any additional heat onto the pump (however minimal that appears to be now) and also give the fluid a little bit longer to cool down in the resevoir before being pumped back into the system.

Again, great info guys, really appreciated.

 

[magoo]

3/8x1/2 inch tubing (what is the importance in tubing size other than reduced kinking with large sizes)

Nothing really. Some people like the larger pipes because they can bend a bit easier, some like to put a big pipe on an even bigger fitting so as not to need fasteners (I just use compression fittings, partly because they look snazzy).

I use 3/8ID 1/2"OD and overcome the bending problems with these: http://www.frozencpu.com/products/6...ls_-_12_OD_Tubing_-_Black.html?tl=g30c289s715 I think primochill makes the best coils, they are easy to work and look really nice.

Here's my tubing of choice, tends to be a bit stiffer and easier to work with : http://www.frozencpu.com/products/1...aboratory_Tubing_-_Black.html?tl=g30c457s1152 I buy mine at Sidewinders, it's 75 cents a foot cheaper!!!

 

[FaRKle0079]

480 radiator with something like these fans ( http://www.frozencpu.com/products/87...?tl=g36c15s562 )

If it's going to be external why not go with something like a MORA?

MCP665 Pump

This pump might be on the low side for running three gpu blocks, radiator, and QD's (you'll definitely want these with an external setup). QD's are terribly restrictive. If you do wish to run this way I would set the GPUs up in series so that they maintain a decent flowrate. There's no way you'll get a good flowrate through the blocks with just one D5 and a parallel GPU setup. Even better would be two D5s or two DDCs.

Personally I'm a fan of 1/2" ID 3/4" OD tubing since it has slightly better flowrates and thicker sidewalls for less kinking.

I also planned on the sequence being pump>blocks>rad>resevoir. My line of thinking is that it wouldnt introduce any additional heat onto the pump (however minimal that appears to be now) and also give the fluid a little bit longer to cool down in the resevoir before being pumped back into the system.

Loop order doesn't matter other than placing the reservoir before the pump. The temps will be the same whether you put the rad or blocks first.

 

[USMCGrunt]

Nothing really. Some people like the larger pipes because they can bend a bit easier, some like to put a big pipe on an even bigger fitting so as not to need fasteners (I just use compression fittings, partly because they look snazzy).

I use 3/8ID 1/2"OD and overcome the bending problems with these: http://www.frozencpu.com/products/6...ls_-_12_OD_Tubing_-_Black.html?tl=g30c289s715 I think primochill makes the best coils, they are easy to work and look really nice.

Here's my tubing of choice, tends to be a bit stiffer and easier to work with : http://www.frozencpu.com/products/1...aboratory_Tubing_-_Black.html?tl=g30c457s1152 I buy mine at Sidewinders, it's 75 cents a foot cheaper!!!

What a coincidence, thats the exact tubing I was looking at and +1 on compression fittings, they're cool lookin indeed.

 

[MD]

@ FaRKle0079

Every radiator has heat capacity rating though mostly they do not come with data sheet. Koolance 400 watt rating was example of bad marketing. I though that was clear, I have said You can't always trust heat rating and I gave example Koolance.

Sometimes you can find graphs for given radiators. Some showing what happens with change of airflow, other showing what happens with change of water flow and one with delta regarding on heat load. Last shows nice were is a overload and amount of heat rad should not be exposed to. Sometime you can find review with graphs and info in practical use with airflow and water flow numbers. Info is mostly there only you have to look for it.

Giving rating without water flow and airflow figures as well as without delta between in and outgoing temps is useless. For PC cooling it is useless if radiator can cool 500 watt at 100 deg C with delta of 2 deg C even if it would not over load it.

So I have simplified and said practical rating of modern high end radiators (supplied with enough water and air) is around 300 to 450 watt per 120 millimeter spot. Take it with grain of salt as it is simplified.
This numbers are true with normal PC loads. PC loads are dynamical and load of let we say CPU that produces 250 watt max will do 180 effectively except when using burn in programs and than it will range 230 to 250 watts. Normal use how ever intense it might be is newer reaching top values of rating. 300 to 450 watt rating for high end radiators is safe assumption when calculating normal PC water cooling loop and you now it.

Used with 20 FPI rad's AP do good.(think Swifttech GTS and GTXetreme)

I use Triebwerk 1200 rpm and they are good. Not best, but get job done.

Regarding EC6 I have had external test setup when I was testing some stuff and hose clamp on exit of a pump failed. 355 with XSPC res-top spilled almost a 1L of EC6 over 3 PC's and whole power grid (all power sockets, all power adapters everything). Let me just mention I'm in Europe hence 230 V in sockets. Test loop was running for day's so EC6 was contaminated with all residue from setup and EC6 went over all dust found on equipment and still nothing burned. If EC6 gained on conductivity it gained so little that it did not do anything to equipment and it have not tripped power brakers/fuses. AlI that was needed was to dismantle few peaces, dry them and put them back together. Hence I have great trust in EC6. Also algae build up is almost 0. EC6 is mild and will not eat up O-rings or corrode your copper as some coolant's will. Great coolant. I do not work for ThermoChill and I do not touch any of their other products.

Strange but here in Europe I have never heard of nickel falling/pealing off from EK. Though I have to say they have had revision of FC6970 and they have changed FC6970 model numbers and now they are also cost like 6 Euro more. Might be they have had issues though I never heard of it. I got 3831109855867 model old model was 3831109835265 if I'm not wrong. Yet again I do not work for EK. I would not recommend all of their products though some of them are working great.

Every pump sucks in. Only some are bad in it and you can not relay on it. That is one of the main reasons why you never go with pump to res like I have said. Little pumps are mostly bad at sucking in. All help is welcome/needed as I have stated. DDC and D5 will work when reservoir is at same level and there are many reservoir/top combos were you do not have any drop what so ever. I personally have tested two different res/tops on top of a loop with dual bay res having effective drop of 1/4 of a inch and DDC was still kicking ass
http://www.xs-pc.com/products/laing-pump-top-reservoir/acrylic-dual-5-25”-reservoir-for-laing-ddc/ .
Also tested with this http://www.xs-pc.com/products/laing-pump-top-reservoir/acrylic-tank-reservoir-for-laing-ddc/ and there you have like 2"drop. On top of system no worries. I used that restop for testing when hose disconnected. Pump was with that res on top of PC and 1L of EC6 went out before I could really react.

If pump does not have to suck air but it can suck in water it will work and perform excellent. If loop is pushed and not sucked from system will work. (sucked from like in disaster scenario pump > resevoir >rad > wb > back to pump and reservoir on bottom)

So to be clear I have never said DDC or D5 is designed to suck in coolant (as in when reservoir would be under a pump and pump would need to suck it up hill) and to work on suction as contrary I have said they suck at it and help them by placing res higher than pump and I have said loop functions on pushing water true loop.

All in all you have posted pretty aggressive replay and quotes without reading good what was said. No need for that.I do have to say English is not my native language and I might have not been that clear. Still you could have asked your self what did he wanted to say and you would understand. I have simplified things to be able to help out as easy as possible. All I have stated I have tested and I stand to it till I see it is wrong. If I was wrong and I see it I have no problem saying I was wrong. I have not said anything wrong.

I have also said some changes in order of a loop are making almost no difference (res > pump> wb > rad or res > pump > rad > wb) and I have said I prefer hence not it is best for every situation or it makes huge difference. I have always stated my preferences not a rule.


Practical knowledge not internet reading knowledge here.


MD

 

[magoo]

Practical knowledge not internet reading knowledge here.

Dude............ever listen to Led Zepplin's "Ramble On" ?

 

[FaRKle0079]

300 to 450 watt rating for high end radiators is safe assumption when calculating normal PC water cooling loop and you now it.

My issue is this, you said the above, and then you recommend OP gets a 360 radiator at minimum. Those two statements are contradictory. OP stated his goal was to keep temps below 100C with low to medium speed fans. I feel your recommendation is akin to telling somebody to buy a Porsche when all he wants to do is cruise at 80km/h. Also, as you stated a CPU that's OC'd and might pull 250W during benching will pull less power during normal operation. The same applies to GPUs so in all reality it's not going to be pulling 250W+/card that often.

Anyhow, this point is moot now that OP's decided to go with an external setup. Before he inferred he was space constrained, hence only wanting to deal with a single or dual radiator. Personally, I agree with you that more rad is better and that's the route I would take, but given OP's requirements, the recommendation didn't fit very well.

Hence I have great trust in EC6.

I've had a DDC shoot water all over the inside of my PC once while on. Nothing shorted, but does that prove distilled water isn't conductive either? If none of the water was actually bridging the contacts then there's nothing to worry about. I don't know what happened in your case since there aren't pics, and I assume "power grid" is the same thing as our "power strip," but it doesn't sound like traces/contacts were bridged.

Strange but here in Europe I have never heard of nickel falling/pealing off from EK. Though I have to say they have had revision of FC6970 and they have changed FC6970 model numbers and now they are also cost like 6 Euro more. Might be they have had issues though I never heard of it. I got 3831109855867 model old model was 3831109835265 if I'm not wrong. Yet again I do not work for EK. I would not recommend all of their products though some of them are working great.

Example A
Example B

The short version: EK blames the customer for all quality issues whether it's cracking acrylic tubes or their shitty nickel process. A quick google search will turn up even more problems.

So to be clear I have never said DDC or D5 is designed to suck in coolant (as in when reservoir would be under a pump and pump would need to suck it up hill) and to work on suction as contrary I have said they suck at it and help them by placing res higher than pump and I have said loop functions on pushing water true loop.

Sounds like language barrier. Both of your previous statements about these pumps make it sound like it could suck in water in an open loop, which is obviously not the case.

 

[USMCGrunt]

Sorry, never meant to indicate that I was going with an internal setup but I also didn't specify. My main goal is to keep the cards cooled to around the same capability as the stock cooling...just not with the deafening roar of their fans. 40-50c idle and preferably under 100c at full load. I think I failed to be specific and being ignorant to water cooling in general doesn't help.

So would a single 480 meet those goals or would a pair of 480s be more realistic?

 

[Tsumi]

Sorry, never meant to indicate that I was going with an internal setup but I also didn't specify. My main goal is to keep the cards cooled to around the same capability as the stock cooling...just not with the deafening roar of their fans. 40-50c idle and preferably under 100c at full load. I think I failed to be specific and being ignorant to water cooling in general doesn't help.

So would a single 480 meet those goals or would a pair of 480s be more realistic?

Single 480 with 1200 RPM fans sounds perfect for silent cooling with 3 6970's. A pair of 480's? I would go with one of those Mo-Ra radiators (3x3, so equivalent to 3 360's). And then maybe you can go passive cooling.

 

[FaRKle0079]

My main goal is to keep the cards cooled to around the same capability as the stock cooling...just not with the deafening roar of their fans. 40-50c idle and preferably under 100c at full load. I think I failed to be specific and being ignorant to water cooling in general doesn't help.

So would a single 480 meet those goals or would a pair of 480s be more realistic?

A single thick 240mm radiator (RX240, GTX240, SR-1 240, X-Changer 240, ect) with mid-speed (~1500rpm) fans will meet your specs. How hot are your ambients? A 360mm radiator will put you close to a 10C delta (what most WCers shoot for), and your cards likely won't break 50C (assuming 25C ambient, stock clocks/factory OC, gaming loads) . A 480 radiator will keep everything below 45C. I can say that with confidence since the stress testing heatload of my two 1015MHz unlocked 6950s at 1.3V and i7 920 @ 4ghz don't cause the cards to push above 45C with ~25C ambients on a TFC 480 radiator. Two 480 radiators are a waste of money.

 

[tangoseal]

Lol at some.of these comments.

My gpu temps are 90f idle and 140f full load. I would do c cept I'm on my cell.

I use one 3 x 120 mm rad for 4 gpus and 1 cpu and it runs those temps all fucking day. Go overkill if you want there are so many wrong suggestions on the internet. I use Swiftech rads because they are a high fin density which is better than those giant ass car rads other people use.

 

[Tsumi]

Lol at some.of these comments.

My gpu temps are 90f idle and 140f full load. I would do c cept I'm on my cell.

I use one 3 x 120 mm rad for 4 gpus and 1 cpu and it runs those temps all fucking day. Go overkill if you want there are so many wrong suggestions on the internet. I use Swiftech rads because they are a high fin density which is better than those giant ass car rads other people use.

Unless you're running on extremely high speed fans, or you have your air conditioning blowing right on your radiator, or your house is just plain cold, I don't see how that is physically possible.

Swiftech does not have high FPI radiators. They're considered midrange. The high FPI radiator designs that I'm familiar with belong to the Black Ice GTS and GTX series. FYI Swiftech has a FPI of 12. The GTS has 32, and the GTX has 25.

 

[FaRKle0079]

Lol at some.of these comments.

My gpu temps are 90f idle and 140f full load. I would do c cept I'm on my cell.

I use one 3 x 120 mm rad for 4 gpus and 1 cpu and it runs those temps all fucking day. Go overkill if you want there are so many wrong suggestions on the internet.

Looks like you need to update your signature then:

Intel I7-2600K - Wet --MCP655/35x pumps / 2x12cm and 3x12cm rads
Asus Maximus IV Extreme
16GB Corsair Vengeance DDR3-1600
Corsair Force 3 120gb OS disk
4 1TB WDFALS Tler enabled/ on adaptec 6805 w/ zmm module
Radeon 6970 (x1) + 6990 crossfire (Both Aqua Cooled)

I count 5x120mm worth of rad and and only three GPU cores. BTW, your temps mean nothing without ambients, nice troll post though.

I use Swiftech rads because they are a high fin density which is better than those giant ass car rads other people use.

You mean it's not because you like their cheap and flakey paint along with their bent barb ports? Yeah, been there, done that with swiftech and not coming back.

 

[Erasmus354]

I just want to jump in and say that MD has posted a lot of good information but a couple of things are misleading and not exactly correct.

1) Serial vs Parallel for the cards. Parallel will be marginally better, but not enough to really matter. In any normal watercooling loop the water is moving fast enough that it doesn't heat up much between components (1 degree C is typical). So you aren't in danger of overheating the last card if you run it serially, the temperatures will just be slightly higher than the first card.

2) Water moving too fast is bad because the radiator can't cool it. Just plain wrong. If the water is moving faster it is correct that the radiator will cool it less, but it is also true that the components in the loop have less time to heat it up. In the end faster is almost always better. The faster the waterflow the better it is at transferring heat from the water to the component (or component to the water). So while the radiator is cooling less each pass through the loop, the water is looping more times (faster flow) and cooling more. Just look at any radiator testing, they dissipate more heat with faster water flow.