Print
I’m taking a standard midrange computer with a Core-i7 930 CPU (stock speed: 2.8GHz) and clocking it up to 4GHz. That’ll up my CPU’s heat output, and I’ll need better cooling. Water-cooling can be quieter and more effective than air, but isn’t necessarily cheap or easy to install.
How much time and money do you need to spend to get good cooling? To answer this question, I’m testing three build options: a basic off-the-shelf liquid-cooling loop (Corsair’s H70), an all-inclusive Swiftech DIY kit, and finally, a custom-built water-cooling setup of my own configuration. How can I get the best performance for the least money, time, and aggravation?
Part of the problem with water-cooling (as opposed to, say, conventional air-cooling) is the depth of the market. At the shallow end are closed-loop systems like Corsair’s H70, which you just slap into your case and go. Next come DIY kits, like Swiftech’s H20-320-EDGE, which include everything you need to water-cool your CPU—pump/reservoir/radiator unit, tubes, fittings, water block, and fluid—but require some assembly. Finally, the completely custom option: All you have to do is figure out the best radiator, pump, reservoir, fluid, tubes, fittings, and water block, then put them all together. Making sense of this last option has long been a source of pride and true geek-cred.
Herein lies the real challenge of this month’s build: Given the widespread availability of off-the-shelf cooling loops and systems, from entry-level to boutique, does building a custom water-cooling loop even make sense these days? We’re gonna find out.
Installing an all-in-one liquid-cooling kit like Corsair’s Hydro H70 is about as easy as water-cooling is ever going to get. It’s a great first step for those looking to push beyond the limitations of air, but who are terrified of the destruction wrought by the unintended marriage of liquid and active electronics.
Thankfully, Cooler Master’s HAF X comes with a 14cm rear exhaust fan by default, but also has mounting holes for 12cm fans—perfect for the fans on my Corsair H70.
You can’t hand-tighten it, but the mounting mechanism for Corsair’s H70 is one of the easiest to use of the many water-cooling blocks we’ve installed. Don’t forget the thermal paste!
In attaching the H70 to the rear of the case, I first removed the 14cm rear exhaust fan that ships with the HAF X—the H70’s radiator and fans mount to that space, which also includes 12cm mounting holes. I arranged the device’s two 2,000rpm fans in a push-pull format. I set up the exterior fan to deliver cool air into the radiator from outside the chassis while its companion, the interior fan, was tasked with accelerating the process by sucking this air through the radiator as fast as possible.
The H70, like many mainstream all-in-one setups, includes a pump built right into the cooler’s CPU water block. After adding thermal paste to the CPU itself, I fit the water block underneath its retention mechanism, and then twisted a screwdriver a few times on each of the four screws that run through the retention mechanism into a backplane behind the motherboard.
Tightening the screw on a water block is an art: Make sure you’re applying enough force to keep the water block firmly adhered to the CPU, but not enough that you’re hurting your motherboard.
To lock the water block into place, I twisted it ever-so-slightly to the left so that notches built into its frame fell underneath the corresponding locking elements on the retention mechanism. I then tightened the retention mechanism into the board by turning its four screws (gently) until I could twist no longer—a tight connection is important, but you should never force it. I then connected the fan cables to the CPU fan header and the pump cable to another 3-pin header on the motherboard.
In my tests, the H70 performed adequately in its cooling benchmarks, drawing an idle temperature of 35.3 C on our 4GHz overclocked system. Kicking the system up to 100 percent CPU use across all four cores brought temperatures to an average of 74.3 C, with a maximum average recorded temperature of 75.5 C. Not too shabby… but not great, either.
If you want the advantages of water-cooling without the huge footprint or mess, a single-radiator all-in-one setup causes minimal disruption.
Opinions vary on whether to install your radiator fans as intake or exhaust. We prefer intake.
Comments are closed on this article
aferrara50
April 25, 2011 at 5:02pm
you fail to detail the difference between radiators and how different radiators are built for fans with varying static pressure. A black ice gtx series rad with apx 1800+ rpm fans will hand every other radiator its butt by comparison and SR-1 series radiators dominate in the sub 800 rpm range. Custom loops are the only way to go since you can tailor it to your specific needs. Swiftech makes decent bang for your buck equipment, but it is far from high end. The water/air delta determines the quality of the loop, not the temp of the cpu since it could be a bad mount. The better system will have a lower temp at the exit of the last rad in the loop (or a generally lower water temp overall). I went from a custom $400 loop for my rig to a $2300 loop and the difference is absurd. 2C water/air delta and the computer is completely silent. The difference that a custom loop brings to the table isn't with budget builds, but rather high end, high tdp builds. Really anything can cool just the cpu (as proved by the H50), but once you start throwing in multiple gpus that will really test the quality of the loop. Cmon guys! this is the 2nd wc build posted on here without using pumptops for the pumps! even last year's dream machine's DDC didn't have a pumptop which increases performance by up to 40%.
syddraf
April 15, 2011 at 2:32pm
so I am currently considering watter cooling for a crossfire/sli build due to the fact that air flow between the two cards won't be that great. my question is what would be the best way to set it up?
should I run more than one radiator or leave it with one?
DarthBeavis
March 19, 2011 at 9:40am
I build more water-cooled systems each year than most people here I bet. If you want an immediate 15c drop in temps use Jager as your coolant.
neural0
July 28, 2011 at 1:19pm
I really wish I could say that I didn't google "jager cooling" right after I read the post...
DarthBeavis
March 19, 2011 at 9:40am
I build more water-cooled systems each year than most people here I bet. If you want an immediate 15c drop in temps use Jager as your coolant.
skinnee
March 17, 2011 at 9:13pm
Wow... this turned in to quite the comments thread. :)
Lots of numbers and suggestions being thrown around and there are several factors that could have lead to this 8C difference in testing of 3/8" tube vs. 1/2" tube. Martin used the estimator, which doesn't have absolute accuracy either, but it does point out the difference in tube size (with all other factors remaining constant) is minor and probably not going to mean squat at the end of the day.
Instead of pulling out the soap box... TheMurph, feel free to contact me in the future if you're looking for some liquid cooling assistance, Skinnee Labs would be glad to help ensure consistent and meaningful testing for MPC readers.
-- Cameron (skinnee) Shears
Martinm210
March 17, 2011 at 5:48pm
Regarding your comment about tubing size is everything:
Using the following components:
Swiftech Apogee XT, MCR-320, 6' of tubing, one reservoir and a D5/mcp655/pmp450 pumps, you would see roughly the following flow rates:
1/2" ID tubing = 1.20 GPM
3/8" ID tubing = 1.19 GPM (-.08C)
A net flow rate loss of 0.01 GPM or 0.8% and a temperature loss of .08C.
Considering the DTS sensor resolution is 1 full degree, your average user could not see a measurable difference between the two tubing sizes.
I'm not sure what occurred to create your 8C difference when switching tubing, but I assure you it is not due to the restriction difference in tubing size when using strong pumps such as the Laing series pumps. It's just not physically possible and I believe you may have experienced some other variable causing the change for you.
I personally use 7/16" x 5/8" OD tubing, but for cosmetic reasons, 3/8" ID is perfectly fine for users and will not make any substatial or even practically measurable difference.
Anyhow, just passing along some scientific data for your consideration.
Cheers!
Martin
TheMurph
March 24, 2011 at 1:16pm
I'll paste our final numbers and configuration, just so we're all on the same page.
Test equipment: Swiftech H20-X20, which consists of an, "Apogee XTL CPU water-block, MCR220-Drive radiator with integrated MCP35X pump and reservoir, fans, Radbox and accessories, 6 ft 1/2" tubing." Run on an Intel Core i7-930 CPU.
Coolant: Distilled water + Swiftech HydrX additive.
All temperatures recorded after an hour has elapsed . Number shown is the average of the recorded temperature (celcius) for all four cores. The two tubing tests were run immediately after each other to negate the effects of ambient temperature.
Swiftech setup with 1/2-inch ID tubing:
CPU Average Temp Idle (no overclock; 1603 GHz frequency recorded): 30.5
CPU Average Temp (100% Burn, all 4 cores, 2940 GHz frequency recorded): 47.5
CPU Maximum Temp recorded (100% Burn): 52.75
CPU Average Temp Idle (overclock enabled; 2280 GHz frequency at idle): 36.5
CPU Average Temp (100% Burn; 3990 frequency): 68
CPU Maximum Temp recorded (100% burn): 69.75
Swiftech setup with 3/8-inch ID tubing:
CPU Average Temp Idle (no overclock; 1603 GHz frequency recorded): 33.75
CPU Average Temp (100% Burn, all 4 cores, 2940 GHz frequency recorded): 55
CPU Maximum Temp recorded (100% Burn): 56
CPU Average Temp Idle (overclock enabled; 2280 GHz frequency at idle): 38
CPU Average Temp (100% Burn; 3990 frequency): 81
CPU Maximum Temp recorded (100% burn): 83.5
JBC548
March 17, 2011 at 3:45pm
Why don't you install the fan and radiator inside a small refrig? That would double its BTU capacity, and you'd never have an overload situation on the CPU. You could use anti-freeze to prevent any possibility of freeze-up.
Shazza
March 17, 2011 at 2:41pm
You state you saw a 13+ degree difference between 3/8" and 1/2" tubing size. You have also brushed off concerns expressed over this difference by explaining this was in a loaded/overclocked situation.
Please, please, please go back and do some retesting. By making this statement about such a large temp difference between the tubing sizes, you have lost all credibility with people who build water cooled setups. And, you are propagating a myth that gives many people misinformation.
Most of us who are into water cooling as a hobby have built multiple systems, trying all types of configurations. And, it's common practice to test our systems at full load. Many people get started in this hobby because we want to overclock - so it's not like thousands of experienced people would have missed this tubing size issue.
Please don't ignore my request. I was delighted to see more coverage of water cooling in Maximum PC, and would like to see even more in the future. BUT, either something else was going on that you didn't notice, or you have run into a one in a million configuration where tubing size is so critical. While I seriously doubt the second, I'm always open to learning something new.
Respectfully submitted.
Waco
March 17, 2011 at 10:49am
I honestly don't know how you managed to create a setup that barely outperforms the H70...and gets beaten by the Swiftech kit!
$500+ and you BARELY beat a $100 cooler. How could you ever possibly think that you did this properly?
The conclusions, data, and hell, everything in this article are so far from the truth that I felt compelled to make a new account to post.
Tube size SHOULD NOT affect your temperatures by more than a couple percent - even going from 1/4" tubing to 5/8" tubing.
I just don't know what to say.
TO EVERYONE: If you're looking into doing watercooling - DO NOT follow any suggestions in this article unless you simply want to waste your money. You can do far better for less money.
fallout330
March 16, 2011 at 4:32pm
You mentioned that you added thermal paste on the CPU to install the Hydro H70. Does this mean there was no thermal pad on the H70 prior to install? When I installed my H50, there was already a thermal pad on the block, so I did not add an additional thermal paste.
Thanks
Brechan
March 15, 2011 at 8:48pm
Dispite what others have said about the article; it gave me an informative view on different ways to 'water cool' a typical computer, though I would have chosen a different setup rather than Swiftech-based on cost alone.
There are more water cooling manufacturers coming out with their own versions of the "all-in-one"; so it is good see that there are choices for those of us that are just getting our feet wet-so to speak- moreover, some choices are less expensive, some more...but at least we have choices now.
I have tried the "closed-loop systems"; namely, the CoolIt ECO A.L.C. and the CoolIt Domino A.L.C., and found that they just didn't cool my system to the temperatures that I thought they would (an AMD Phenom 9950 idling @ 39c). This is why I have chosen to go the route of true water cooling.
If the MPC gods would smile apon you, it would a good read if you could compare 3 different water cooling kits from Swiftech, XSPC, and EK Waterblocks.
Elfie
March 15, 2011 at 6:15pm
I've been using 3/8" tubing for a year but now this has inspired me to replace all my bitspower compressions with the 1/2" version to get such a massive drop in temps. Although the fittings will cost the equivalent of $110 here in the UK it's going to be well worth it for such an improvement. Thank you for this helpful article.
TRoiKA
March 16, 2011 at 1:37pm
I wouldnt make that switch so fast. I would check out other sources of information and other tests before spending that kind of money to possibly be dissapointed. I can tell you that a 14 degree drop in temp for a 1/8th inch difference in tubing is NOT normal. I would just advise you to do more research before making that type of purchase!
D00dlavy
March 15, 2011 at 10:34am
iBuyPower puts water cooling on their systems for free. My pump was so loud, though, plus the hoses were long and they were sort of kinking - I just took it off and installed a really large, quiet fan and heatsink.
I forgot how much force it takes to install heatsinks on AMD processors.
klue22
March 14, 2011 at 10:03pm
Lordy 70 degrees at full burn? I can't even hardly break 55 on my 970 at 4.2 non-turbo. Granted its a 32nm arch but at the same time I'm also running 2x 285s in series in the same loop. I have an additional rad as well but your temps still seem hot because I can't even begin to approach them even with the CPU and GPUs folding 100%.
MattyMattMatt
March 14, 2011 at 7:11pm
Where do you find an Antec TPG 850 for 125? They're sold out where I shop and 150 or more :(
silent975x
March 14, 2011 at 4:19pm
There is so much misinformation in this article and in these comments, it's disgusting.
Anyone who wants legitimate information about liquid cooling should go to www.xtremesystems.org/forums
MaxPC has always been so fails at enthusiast cooling systems - give up, or get serious.
Stvwndr219
March 15, 2011 at 6:48am
Here's an interesting thread about the testing methodology:
http://www.xtremesystems.org/forums/showthread.php?t=267696
And both Cathar and Martinm, both respected voices in the W/C community point to there being a negligible difference betweeen 3/8" and 1/2" tubing. Cathar's test is below:
http://www.xtremesystems.org/forums/showthread.php?t=147767
This was also done with a full load on the computer. Regardless, a 10C difference due to only tubing is absolutely absurd. Can you verify that both mounts were exactly the same, both times the seating was correct, and the correct amount of TIM was used? All of these factors would contribute to that 10C difference but tubing certainly would not.
silent975x
March 14, 2011 at 7:29pm
First off, yes the D5 (MCP655) has more flow rate than something like a DDC (MCP35X). However, this is COMPLETELY offset by that mass amount of quick disconnects. The 655 has far less available head pressure compared to the 35X - pressure you will surely need when trying to push water through 6 quickdisconnects. Speaking of which, that many QDS will be about equivalent to adding the restriction of another CPU block into your loop.
Second, 3/8" tubing compared to 1/2" tubing has been shown over and over again to have negligible difference. In fact, splitting the difference with 7/16" ID tubing can often work out better because you have that negligible diameter increase, whilst also having the flexibility of 3/8" for tighter tube routing. You must have the only test results which show 1/2" ID to be far superior - might make you question your testing methodology?
Third, $558 for a CPU only loop? My god man. About twice what you needed to spend. Of course it wasn't helped by the sinful amount of QD bling you stuffed onto every component...but whatever.
P.S. Tygon is expensive, and really isn't better than Primochill Pro LRT which is far easier to work with, and looks nicer ;)
Good choice on the fans though? I saw someone below dissing on them, more fails.
TheMurph
March 14, 2011 at 11:49pm
"You must have the only test results which show 1/2" ID to be far superior - might make you question your testing methodology?"
No. As mentioned below, there is a negligible difference for tube sizes at lower CPU temperatures -- e.g. running our processor idle on 3/8 or 1/2 didn't show much of a difference at all. The difference became far more apparent, however, the more one generates a crap-ton of heat for the rig--in this case, kicking the processor up to 4.0 GHz and letting it rip.
$558 is expensive, yes. I happen to think that the Quick Disconnects are worth every penny. And I'd like to see the numbers you're using when you bring up flow restriction in the quickdisconnects. I quote:
"In the end, I can certainly say I was wrong on Quick Disconnects. They do not bring the restriction I thought they did to our loops, not sure what I was thinking or if I was just over concerned about it."
--http://skinneelabs.com/quick-disconnect-roundup/
silent975x
March 15, 2011 at 1:29pm
"In the end, I can certainly say I was wrong on Quick Disconnects. They do not bring the restriction I thought they did to our loops, not sure what I was thinking or if I was just over concerned about it."
Didn't bring the restriction he thought, does not mean they did not bring restriction. Careful with your quotes.
Look at skinnees graph comparing a vanilla D5 loop to a D5 loop with only 2 QDS. The 2 QDS (V3LN) dropped flow by .19GPM from the vanilla 1.61GPM. Now, assume a linear path, and add 2 more pairs of QDS into that loop, dropping it by an additional .38GPM. Now you have 1.04GPM for a total drop of .57GPM - about a 35% reduction in flow from the base loop.
Can i ask you a question? How often do you expect to be changing/cleaning your loop that you need a QD on every component?
Not going into the tubing debate, because I see someone else has spoken on that ^
FuzzDad
March 14, 2011 at 3:46pm
What was the ambient temp in the room? I ask because the CPU temps seem a tad high at idle and I think you *should* have gotten better temps at full power w/the more traditional water cooled systems. Over 70C just seems too high for those setups. Maybe it's all the quick disconnects...they do reduce flow to a point.
I also agree w/you on Koolance products...their new CPU and GPU waterblocks are excellent (as are their fittings...the 3 series seems to grade higher on spillage and flow as opposed to the newer 4 series but both are solid disconnects)...but I have to say I've never seen that many quick disconnects on a system. Unique it is!
Lt.Nefarious
March 18, 2011 at 2:10pm
I'm running a Core i7 920 at 4.0 GHz (1.4 GHz overclock) with a full EK motherboard water block and a EK HF supreme water block running distilled water will a silver kill coil my temps are 38 C idle and 49 C at load. Something is definitely wrong with the setup!
FuzzDad
March 14, 2011 at 6:44pm
Maybe not...what were your OC specs (vcore, etc)? I have a i7 920 w/dual 460's in a single loop w/360 and a 120rad and at 3.8gHZ I'm in the low 60's on a EK Supreme HF w/EK 12V pump (on distilled). Tubing is 3/8".
TheMurph
March 14, 2011 at 6:56pm
Hm. Surprised MPC didn't print the sidebar on overclocking that went with this article in the print mag. Here's a rough draft of it:
So you want to jack your processor up to insane speeds, don’t you? Alright. Here’s how I overclocked my Core i7-930 test system in a nutshell: Cranking up the speeds on a Core i7 chip requires one to head into the BIOS and up the CPU’s base clock from its default of 133 to whatever number you think your system can handle. The base clock times the CPU’s multiplier—21, in this case—equals the processor’s final, overall speed. I had to jump mine to 190 to hit the four-gigahertz bar.
But! The base clock directly affects the speeds at which both the system’s memory and Quick Path Interconnect will run. Pushing either too high (above my RAM’s rated value of 1600 MHz or the QPI’s 4.8 GT/s default) leads to system instability slash death, so I had to manually set the former to 1,523 MHz and the latter to 6.85 GT/s within the BIOS.
Not done yet, though! The hungry system’s components demanded more energy for stability, so I had to kick up the CPU’s voltage to 1.4, the QPI’s to 1.4, and the RAM’s to 1.64—all of .01 below its rated voltage limits. Now that’s hardcore—hardcore, stable, and fast!
FuzzDad
March 14, 2011 at 8:57pm
Ok...i have ioh 1.2, ich 1.2, dimm at 1.65, cpu pll 1.88, qpi/uncore 1.35, vcore 1.4Bclk w/191 to hit 4MHz. Using Prime95 I hit a max of 64C on two of the four cores (after 45min) and as I type this (10min later), my idle temp is 33/34C. What's your TIM?...that may be some of it...I picked up Indigo Extreme and it dropped my hi-end temps 2-3C from Arctic 5. Regardless...being able to run these processors at 4MHz w/o really worrying about burning the house down is a big plus.
One thing's for sure...if you go into water cooling the money leaps out of your wallet because as soon as you try it you want to buy more and more equipment, upgrade to this, install that, change your reservoir/pump/whatchamacallit/etc...it's like custom car work...and $$$
TheMurph
March 15, 2011 at 12:00am
I presume you've also kicked off turbo boost and speedstep?
fwiw, my overclock temps hover around those MPC generated in its initial article for an i7-920:
Of course, the 80 degrees or so was done with an air cooler. I would expect a water-cooled rig to be up there, but not quite that high.
Also, fwiw, all testing for this article was done in the exact same location with the exact same procedures (in terms of keeping the room temperature as similar as possible). It was a bit warm in here, I'll give it that -- but that warmth would have affected each and every tests, making it a moot point.
FuzzDad
March 15, 2011 at 11:20am
Yup (for the test)...I like leaving them on however because i leave my PC on for long periods of time and it's nice to let it idle at 2MHz rather than keeping things hot all the time. I use the OC info listed here for mine: http://www.overclock.net/intel-cpus/538439-guide-overclocking-core-i7-920-930-a.html
What's interesting is watercooling developed into a very high-end niche but the inclusion of the H50 brought thousands of neophytes into that world...and after using the H50 people w/enough disposable income are trying more hardcore approaches. You even see manufacturers mixing it up w/OC'rs while they build their new products in several of the more lively forums based on water cooling and overclocking. I even saw a small firm posting images of the water block they were milling. It's pretty "cool". lol.
TRoiKA
March 14, 2011 at 2:15pm
In the article "how to build a kick ass liquid cooling system - 6 simple steps" you guys mention that "the demonstratable difference in between the two sizes of tubing (3/8 and 1/2) was slim" but now you say it drops the temps TEN degrees? You even suggested choosing 3/8 due to its slimmer tendency to bend! Also, in that article, you reference the fact that all "modern" cooling systems use some kind of anti conductive and anti corrosive coolant, whereas here you denounce these fluids like they are the biggest waste of money. Im planning on water cooling soon but all these confliciting views are confusing me! Can someone provide clarification?
TheMurph
March 14, 2011 at 6:39pm
I can't speak to the other article; I can only speak to this one. I would indeed add some kind of anti-microbial agent to distilled water if I was to use it in my rig. My point in this article was that one does not need to spend $50 on the super-fancy fluids when distilled + an additive will work just as well.
As for the tubing, you won't see as much of a difference in temperatures on an idle CPU. When or if you start hitting full burn, however, half-inch tubing will allow you to keep your temperatures lower.
Danthrax66
March 14, 2011 at 2:54pm
They did something wrong if you want real liquid cooling help you should look at other sources evga forums is one of the better places. But yeah the general consensus is that 3/8 and 1/2 inch doesn't make that much of a difference but if you can go 1/2 then by all means go for it.
Here is a very informative list of stickies the first link doesn't work though http://www.evga.com/forums/tm.aspx?m=574023
TheMurph
March 14, 2011 at 6:36pm
In our tests, using a 3/8-inch tube instead of a 1/2-inch tube resulted in a 13.75-degree difference on our overclocked rig. That's 13.75 degrees hotter, mind you.
We didn't see nearly as much of a difference on a stock clock or overclocked rig running idle--as expected.
Danthrax66
March 14, 2011 at 12:30pm
This is the worst custom liquid loop I have seen. Koolace??? Are you serious Koolance makes shit.
http://www.sidewindercomputers.com/ekwaekhfac.html <-- better water block
http://www.frozencpu.com/products/11110/ex-tub-703/Bitspower_Matte_Black_Ultimate_G_14_Thread_12_ID_x_34_OD_Rotary_Compression_Fitting_BP-MBRCPF-CC5.html?tl=g30c409s1033 <-- better compression fittings
Why would you use those fans??? Those aren't that good for lquid cooling you need fans with high static pressure.
Oh and where did you get those outrageous prices? http://www.sidewindercomputers.com/swmc12vdcpuw.html <--$67 for the pump
Also that radiator isn't adding much you would be better off with one of these http://www.frozencpu.com/products/5069/ex-rad-93/Black_Ice_GT_Stealth_360_X-Flow_Radiator_-_Black.html?tl=g30c95s161 cheaper and roughly equal performace for your needs.
Also $70 for a tank? http://www.frozencpu.com/products/9577/ex-res-158/Bitspower_Water_Tank_Z-Multi_400_Inline_Reservoir_BP-WTZM400P-BK.html?tl=g30c97s165 <-- cheaper
And don't tell people to just run straight distilled water or else they will end up with a tank full of aquatic life and a few ruined blocks and radiators. You sould probably be using a silver kill coil http://www.petrastechshop.com/sikibyia.html and some PT Nuke for anticorrosion http://search.store.yahoo.net/yhst-65556269779593/cgi-bin/nsearch <-- you only need one of those either copper based or PHN based depending on the metals in your loop. And probably the most important thing when building a loop: DON'T MIX METALS
TheMurph
March 14, 2011 at 6:54pm
I'm not going to address all of your points, as I don't have time to pull out all of the benchmark numbers. Suffice, you're armchairing your water-cooling analysis based on things you've read. I'm reporting based on things I've tested.
In this case, I indeed used the exact EK Waterblocks product you refer to (the EK Supreme HF Universal CPU Liquid Cooling Block - Rev 2) during my testing. I found that the Koolance waterblocked used in the feature (a new design, at the time) gave us an average temperature of 61 degrees on the full-burn, overclocked test system. The EK block hit 65 degrees.
Also, your suggestion for compression fittings is right out. Please check out the little addendum we did on the Koolance quick disconnects -- I would never use a compression fitting over these. No way, no how.
I'm not a fan expert by any means, but I've done my research and plenty of tests seemed to indicate that the Gentle Typhoons were the way to go based a ratio of noise to cooling potential. I did test different fans myself, and felt that these were the best selection (albeit leading to a little bit warmer of a system) based on the lower noise they generated.
Anyway, I appreciate your comment. Indeed, one can make water-cooling loops in about 100 different ways, as a commenter notes.
Danthrax66
March 16, 2011 at 8:11pm
What plate did you use with the EK block? Also did you try the Heatkiller v3 ? And the biggest problem I have is the inflated prices you used. Also The gentle typhoons really aren't that great due to the low static pressure and it is always easier to get higher power fans and to then use a speed control than to have low powered fans and want to push your system higher. I would really like to see an all out water loop or a custom water loop. You should also factor in the ease with which you can add motherboard blocks and gpu blocks.
Also I;m still hugn up on the results you recieved using the EK HF the Koolance and the Swiftech... I know for a fact that the EK HF beats the Swiftech in temps, yet you are saying that the swiftech beats the Koolance and that the Koolance beats the EK???? Something was seriously wrong with your tests. Maybe you got a bent EK block? Or a misformed one? You guys should try lapping it and see what happens.
AmTGMan
March 14, 2011 at 3:13pm
Actually, Koolance has been making good quality parts for a few years now, rather than the shit they used to push. You just need to know which components to look for. (i.e.: avoid the legacy parts, and go for the newer parts, and ask around on various forums.) The CPU-370 looks be a solid block, if not the absolute best. And switching to a EK Supreme HF (Also an excellent block) would result in a negligible difference.
Also, the radiator you linked really isn't that good. It's cheap, and thin, so it's easy to stuff into a case. However, it has a fins-per-inch count of 30. Most pc radiators have around 8-10 fpi. The reason for that is the greater the fin density, the harder it is for the fans to force air through the radiator. To get enough airflow to take advantage of the high fin density, you need high-rpm fans, and they are loud, removing one of the main advantages of water cooling. (If you need a high fpi radiator, a Black Ice GTX Xtreme would be a better buy, anyway.) Hardware Labs do make good radiators, the one you linked isn't really one of them.
The Gentle Typhoons they used are actually some of the best fans you can buy for radiators, being low noise, yet having plenty of static pressure for a radiator.
The XSPC rad they used is an excellent radiator, as well. It's well suited for low-rpm fans, but isn't completely useless with higher rpm fans. They made a good choice here.
They did go overboard with the reservoir, I'll agree with you there. A smaller one would be a better buy.
While I personally diasgree with how many quick disconnects they used, they should be fine. The ones you linked to are definitely a good choice, though.
And you are correct, straight distilled = BAD IDEA. They should have mentioned the use of a biocide.
Sources:
www.xtremesystems.org
www.overclock.net
www.skinneelabs.com
Danthrax66
March 16, 2011 at 8:16pm
Honestly the radiator wouldn't matter that much you and it would cost less. Also noise doesn't bother me so I would have put some deltas on that bitch and called it a day :D
TheMurph
March 14, 2011 at 6:59pm
"And you are correct, straight distilled = BAD IDEA. They should have mentioned the use of a biocide."
That is correct. I was making the point that distilled water -- as a base -- funcitoned no worse than the $50 stuff one can otherwise buy. Obviously, one would want to pour an additive in there, as we do in the custom system itself. I could have definitely made this more clear, however.
Danthrax66
March 16, 2011 at 8:12pm
Straight distilled water should always function better than the fancy fluids. The fancy fluids sacrifice cooling power for the avoidance of corrosion and aquatic life.
bpstone
March 14, 2011 at 1:49pm
That is the beautiful thing about custom water cooling. There are a hundred different combinations you can use to achieve great results.
Danthrax66
March 14, 2011 at 2:28pm
Yeah and I'm pointing out the fact that they overpaid for a terrible combination lol. I mean I think they would try to at least put some effort forward if they wanted to do a true comparison. But this just looks like an advertisement for Corsair and Swiftech. Although that swiftech set is very good.
TheMurph
March 14, 2011 at 7:04pm
Dude, what are you talking about?
The point of this build-it, as noted in the text, is that we're comparing a custom-build system to all-in-one setups. This is not a value judgment that Corsair or Swiftech represent the best-in-class kits for an all-in-one setups. They are merely test examples that we're using in the "built it yourself or buy it aftermarket" sense of the story.
In a perfect world, I'd love to put 5 different AIO setups through their paces. That was not the purpose of this feature. There might very well be better kits than either company's; I haven't tested that, so I can't say that as a definitive statement myself.
And I put a ton of effort into this piece, I'd like to note -- for length, we didn't get into all the different permutations of loops and products I tested. Perhaps I can explore this in a future column, provided the MPC Gods are feeling generous. ; )
Danthrax66
March 16, 2011 at 8:14pm
But the thing is you paid $500 for a worse performing custom kit? I don't understand how you could have accomplished that...
immersive
March 14, 2011 at 12:29pm
For the life of me I can not figure out what the blue wires are. Are they USB 3.0 from back of mobo to the front headers?
Jarol
March 14, 2011 at 2:07pm
USB 3.0. They come with the case. They go from the very front panel (which is at the top front) to the back. The HAF is suppose to have a ton of room in the back panel for those things to hide, but they did a quick job on it and it shows.
Anyway just popping in to say that I bought that same case for exactly this purpose. The case is built exclusively for watercooling in mind. I too would like to dive in on it, but it is not going to come cheap thats for sure. This article should make sure to name it specifically as the HAF 942 case. There are 3 or 4 models out there and the 942 is the latest one.
