Overclocking

As cooling solutions do a better job of keeping the CPU at a lower temperature, it is reasonable to expect the overclocking capabilities of the CPU will increase. In each test of a cooler we measure the highest stable overclock of a standard X6800 processor under the following conditions:

CPU Multiplier: 14x (Stock 11x)
CPU voltage: 1.5875V
FSB Voltage: 1.30V
Memory Voltage: 2.20V
nForce SPP Voltage: 1.35V
nForce MCP Voltage: 1.7V
HT nForce SPP <-> MCP: Auto

Memory is set to Auto timings on the 680i and memory speed is linked to the FSB for the overclocking tests. This removes memory as any kind of impediment to the maximum stable overclock. Linked settings on the 680i are a 1066FSB to a DDR2 memory speed of DDR2-800. As FSB is raised the linked memory speed increases in proportion. The same processor is used in all cooling tests to ensure comparable results.

Highest Stable Overclock (MHz)

The Swiftech H2O-120 Compact and Corsair Nautilus 500 both topped out at 3870 MHz - near the top of tested air coolers but below the 3.94 GHz of the Thermalright Ultra-120 eXtreme, and the 3.90GHz of five tested air coolers. Once again test results - this time in overclocking - are near the top but do not match or exceed the best air coolers tested at AnandTech.

It was quite a surprise to find overclocks on the P35 motherboard also topped out at 3.87 GHz, particularly since we could not duplicate the 14X base multiplier on the ASUS P5K Deluxe motherboard. We achieved 3.87 GHz by choosing the 333 strap and setting a FSB of 352 with the 11X multiplier. This turned out to be the highest stable overclock on the P35 board with either water cooler - the same results found on the EVGA 680i board.

As stated many times, the overclocking abilities of the CPU will vary at the top, depending on the CPU. This particular CPU does higher FSB speeds than any X6800 we have tested, but the 3.9GHz top speed with the Tuniq is pretty average among the X6800 processors we have tested with Tuniq cooling. A few of the other processors tested with the best air coolers reach just over 4.0GHz, but the range has been 3.8 to 4.0GHz. Stock cooling generally tops out 200 to 400 MHz lower, depending on the CPU, on the processors tested in our lab.

Scaling of Cooling Performance Final Words
Comments Locked

58 Comments

View All Comments

  • geok1ng - Wednesday, July 23, 2008 - link

    There is no reason to dump $120+ on a budget water cooling system for CPUs since most sub$50 air coolers will do a good job

    but with SLI/crossfire systems the typical air cooling solution are suboptimal to say the least. Even you have a board that can fit 2 VGA and 2 Acellero S1s, there is no way to put good airflow averthe humungous setup!

    Someone plaese give us a water cooling solution for multi GPU systems that doesnt cost MORE than the VGAs!
  • Cregaune - Tuesday, March 25, 2008 - link

    This review has given me lots of food for thought. Frankly, I find the conclusion incredible.

    Perhaps you guys are right, but before people jump to your conclusions I wonder why your ambient room temps were kept within such loose margins i.e 20'C to 22'C.........a full 3'C !!!

    Controlling ambient room temperature is particularly important when reviewing a water cooled system, simply because it can take an hour or more for the coolant temperature to stabilise; before you can legitimately take a reading. If, in the meantime, you let the room temperature fluctuate within such grossly wide margins, your results are next to useless....especially given the relative performance figures in your findings.

    Just a thought!


  • Cableaddict - Friday, April 11, 2008 - link

    Wesley,

    As far as performance goes, I'd say you present a valid argument. The specific kits you tested are not worth investing, vs a few of the best air-coolers. So far, so good, but your conclusions and general tone seem to fail badly after that one true point.

    You wrote, (in this thread) "We state you will likely need to spend more than $300 to possibly outperform current top air coolers."

    Nonsense. As many have mentioned, the big problem with these kits is the use of a single "220" size radiator. Using a dual-rad, or two singles, has been shown in test after test to increase cooling dramatically (even with a "lowly" Swiftech rad) So, one could buy the H2O-120 and add another radiator for $42 more. That's WELL under $200, and the performance OC'ed & under load will slightly outperform the Ultra-120. AND , it will be much less noisy.

    You conclusions about noise are correct, but only at lower fan speeds.You posted a chart showing high OC / high load performance, but failed to include noise specs under those conditions. The noise comparisons at lower cooling settings do not apply at all. I've been looking at TONS of tests, specifically weighing cooling vs noise. At the high speeds needed to make the Ultra-120 types perform their best (about equal to a low-end dual radiator WC) they are much, MUCH noisier than WC, according to every test I've read.


    Last: Bear in mind that many folks cannot fit those big air-coolers, espially HTPC people, or those with rack-mount music & video pc's. For such rigs, the best air coolers that will actually fit are the Gemini II, the DuOrb, etc. Even a single-rad WC setup does better than those when OC'ed & under load. (according to many tests I've seen.) With a dual-rad, WC systems decimate these air-coolers. For instance, one test put a Gemini II / dual Noctua blower against the H2O-220 kit. At 3.7 Ghz, under load, the WC was 15º cooler.

    -----------

    Regarding pump noise: Well again, your comparisons are fair if one is going to strictly obey the "rules" and not mod any system at all. However, One can pretty easily add sonex or other noise absorbtion to the inside of the case, and limit a good bit of that pump noise.

    Going to a non-kit system, one can now choose the Lang D5 vario, dial it down a little, and add some sound isolation around it. Not so hard, and all reports say it is then virtually silent. With a dual-rad performance should still be excellent. Granted, that's approaching the $300 mark, but serious overclockers would certainly do better with such a set-up, and it would be LESS noisy.

    I guess one last question must be addressed, and that is: Just how cool do we need to go? Let's say one want to OC a Q6600 (timeframe of this review) to 3.8 ghz. A $75 air-cooled rig will run under heavy load at (let's say) 60º, and a decent WC rig will run it at 50º. Is 60º safe enough, day after day, that the WC system is not worth the substantial cost? I don't know the answer. I wish I did.

    -And when pushing the FSB really hard, one might worry about the rest of the mobo components. -So now we have to weight the WC advantage of being able to also cool the NB and GPU, vs the air-cooler advantage of cooling the ram,NB, mosfets, etc.
    One might reasonably say that,with heavy OC, a WC set-up requires an additional case fan, and that means a little more noise.

    Complicated...
  • Cableaddict - Friday, April 11, 2008 - link

    Custom H2O-220 type system:

    Swiftech Apogee Drive 350 Pump / CPU water block combo $80

    Swiftech MCR220-res: $55

    2 Noctua fans: $50

    non-evaporate hoses / clamps: ~ $20 MAX.

    Hydrex: $3
    ------------------------------------------
    $210 or less.

    Swap-in a Lang Vario D5 pump and an Apogee or Fuzion waterblock, instead of the Apogee Drive 350, and you still under $275.

    Test either of these systems on a Q6600 @ 3.8 Ghz, against any air system, includng the Ultra-120, and then get back to me.

    Also bear in mind that either of these systems will easily also cool a GPU, with another $50 of parts.

  • Cableaddict - Saturday, April 26, 2008 - link

    One more noise consideration:

    Granted this is a review of the H2O single, but for very little more money one can get the H2O with a double rad. THAT system should slightly outcool even the ultra-120, but what;s most important is that it will be MUCH quieter.

    Again, the Ultra-120 types require pretty high CFM's to give good results. A dual radiator can deliver excellent performance with a pair of Nocyua 1300 rpm fans. That's bloody silent.
  • perfectdarth - Sunday, December 30, 2007 - link

    you cant say say that watercooling is bad because a 2 budget kits get beaten by an ultra 120. watercooling does cost more but radiators and pumps last years and for $300 (no more) you could make a custom kit that would be quite a bit better than the top air coolers when cooling an overclocked cpu.

    you harped on how watercooling costs more and takes more time/effort but you didnt state that people that are willing to spend the extra time/money dont always do it just for better temperatures and might find it fun and a bit of a challenge the first time. putting on a heatsink isnt very satisfying, you could train a monkey to do that. you could say people who spend $800 on a graphics card to play games at max are stupid because its still the same game at medium on a $350 graphics card but if you like to lay games that look nice you would think its worth it, it was like you had your mind made up before you wrote this article

    people saying that the hassle of watercooling isnt worth it is just plain ignorance, some people enjoy maintaning their car themselves but other people would go meh and go pay someone to check their oil and water. that doesnt mean that people who dont care should tell the people that do it themselves that they are wasting time or the they should tell the lazy ones they shouldnt be lazy. watercooling is a hobby just as much as it is a way to better cool your computer.

    i would like to see the swiftech h20-220 Apex Ultra tested against the top aircoolers since it is a very good kit and under $300 afaik. also watercooling seems to make a much bigger difference to graphics cards than cpus.

    saying that you would rather buy a phase change over a good watercooling kit is rediculous. phase is more work and just as if not more risky than watercooling because of condensation.

    also why farcry and not use tat or orthos? i didnt like this article much since it is very bias towards air. i read "water is useless air is so much better" all the way through.
  • rabident - Tuesday, September 25, 2007 - link

    You could argue heatpipe coolers are closer to water coolers than air coolers. They're basically sealed water coolers. Most use distilled water as the heat transfer medium, just like conventional after coolers, except heat pipes are more efficient because they involve a phase change (which absorbs a lot of heat).

    The real value with conventional watercooling is the freedom to piece together a custom cooling system that meets your needs. It lets you balance price, cooling performance, and noise the way you see fit. Plus there's a lot of room for creativity.

    I think it's going be hard for water to beat air in price/performance, but if price is the #1 factor then I think the value of any after market coolers is questionable.
  • Starglider - Sunday, September 23, 2007 - link

    This review is ok, nothing amazing but the accusations being thrown around seem unwarranted. However the basic conclusion is this; modern heatsinks are good enough that for a single medium-speed 120mm fan, cooling is airflow and radiator surface limited. With heatpipes and improved layout heatsink thermal conductivity is no longer an issue (for quality air coolers).

    It would be nice if this was made more explict in the conclusions, and I am dubious about the claim that you can't get a better radiator solution (e.g. dual-120mm or reserator) for under $200. The notion that water cooling won't give you any advantage other than chipset/GPU cooling is false in the general case; it is only true if you have a single 120mm fan radiator (plus it ignores the benefit of lowing case internal temps).
  • strikeback03 - Wednesday, September 19, 2007 - link

    So is the "water additive" really just old-school green automotive coolant, i.e. ethylene glycol? Or does it just look that way?
  • aigomorla - Tuesday, September 18, 2007 - link

    http://www.vr-zone.com/articles/Swiftech_H20-120_C...">http://www.vr-zone.com/articles/Swiftec..._Compact...

    Sorry. This review was better writen. And shows this cooler clearly owning air.

    The editor i feel now has a grudge against water cooling. He is pushing air is greater then water when it looks like he has no experience with a decient water setup.

    The pump making a lot of noise? Okey, can one owner who has a DDC confirm this? Because i have 4 and none of them make loud noises.

    The Liang D5 doesnt even make loud noise, and its a bigger pump.

    Sorry your 300+ watercooling comment needed to beat air... thats absolute BS.

    People please do some research b4 you buy into this review. I asked the editor polietly to remove that statement and he has yet to remove it. I ask you guys to check out our case and cooling forum. There is a post on this and listen to what all the members on AT are saying. Only the editot of this article believes in his 300+ needed to beat air. The rest of the members on board totally disagree with him. XS is laughing at anandtech at the moment. Even the air cooling people at XS are laughing because its a KNOWN FACT a 120x2 radiator would demolish anything AIR can bring up. Even TEC air coolers dont stand a chance.

Log in

Don't have an account? Sign up now