AMD and Intel Heatsink Roundup 2006

Cooling/CPU Cooling by jmke @ 2006-12-04

We compare a whole bunch of CPU coolers which fit on AMD/Intel latest platforms and compare them on two hot running systems; which one will keep your CPU the coolest, how quiet can it get? Find out in this last HSF roundup of 2006.

  • next

Introduction & Athlon 64 Test Setup

Madshrimps (c)


Introduction

Today we present you all the heatsink results we gathered in the last year, tested on our Athlon 64 setup, the complete list consists of 36 units which were compared to each other and the stock aluminum AMD heatsink.

Second part of the roundup consists of inaugurating our new Intel S775 test setup, we do this by re-testing the acclaimed Scythe Ninja and compare it to 6 newcomers.


37 Heatsinks Compared

These were the heatsinks tested on the Athlon 64 setup:

Madshrimps (c)


For those having trouble identifying the collection of copper/aluminum above we made this list, if you want more information about a particular heatsink please visit our HSF Overview.

  • AMD Stock Cooling
  • AMD 4 Heat pipe Stock cooling

  • Aerocool Dominator
  • Arctic Cooling Alpine 64
  • Arctic Cooling Freezer 64
  • Arctic Cooling Freezer 64 Pro
  • Asus Silent Square
  • Asus X-Mars
  • Coolermaster Susurro
  • MACS Triumph TEC
  • Noctua NH-U 12
  • Noctua NH-U 9
  • Sharkoon Red Shock
  • Silverstone NT02
  • Silverstone NT06
  • Scythe Infinity
  • Scythe Katana CU
  • Scythe Mine
  • Scythe NCU-2000
  • Scythe Ninja
  • Scythe Samurai Z
  • Scythe Shogun
  • Spire DiamondCool II
  • Spire VertiCool II
  • Thermalright XP-90C
  • Thermalright XP-90
  • Thermalright XP-120
  • Thermalright HR-01
  • Thermalright SI-120
  • Thermalright Ultra-90
  • Thermalright Ultra-120
  • Thermaltake Big Typhoon
  • Thermaltake Mini Typhoon
  • Titan Tower TTC-NK32TZ
  • TTIC NPH
  • Tuniq Tower 120
  • Zalman CNPS9500CU


    AMD Test Setup and Methodology

    All heatsinks except one (MACS TEC) were tested with the case in the exact same environment, with ambient noise levels consistent. The MACS noise level was tested with higher ambient and at a different angle/distance. It can not be directly compared to the other results, and thus its noise levels will be left out the comparison chart on the next page.

    Athlon 64 Setup

    Madshrimps (c)
    CPU Athlon 64 3200+ @ 2400 @ 1.7v vcore
    Mainboard DFI NF3 Lanparty
    Memory 1 * 256Mb PC3700 OCZ
    Other
  • Antec Sonata II AcoustiFan DustPROOF 120mm @ 5v in the rear as outtake (mounted with soft-mounts)
  • ATI R9000 Passive Cooling
  • Silverstone EFN-300 300W Passive Cooled PSU
  • Maxtor 120GB IDE HDD


  • in-take temperature was measured at 22°C for all tests, but temp fluctuations, different mounting and user error can account up to 1-3°C of inaccuracy in the obtained results. Please keep this in mind when looking at the results. Each heatsink was tested repeatedly; if we got questionable results the test was restarted.
  • Noise level of each HSF combo was recorded with SmartSensor SL4001A, the sensor was placed ~50cm away from the case. The lowest dBA reading in the test room was 32.5dBA with everything turned off! Noise produced with system running without HSF fan: 34dBA @ 50cm
  • System was stressed by running K7 CPU Burn for 30min (after Thermal Compound's burn-in); this application pushes the temperature higher then any other application or game we've yet encountered. Speedfan was used to log maximum obtained temperatures.
  • Arctic Silver kindly send us their ?Lumière? thermal testing compound which has the same colour as Ceramique, but only a break in time of 30min!
  • Arctic Silver's ArctiClean was used to clean off thermal paste of the CPU and heatsink between tests

    Fans used for comparison

    To eliminate as much variables in the tests we test each heatsink with a "reference" fan if it can be mounted.

  • Delta NFB0912L 92mm: 42CFM
  • Papst 120mm 4412 F/2GLL: 40CFM

    CPU Speed and Noise Measurement

    Since we’re only using an Athlon 3200+, it would be interesting to overclock the CPU so its maximum heat output increases and it can simulate a higher clocked Athlon 64.

    We recently purchased a power meter similar to this. Doing a few basic measurements with the test system gave these results for full system wattage usage.

  • Athlon S754 3200+ @ 2200Mhz - 1.5v: idle: 67Watt / Load: 125Watt
  • Athlon S754 3200+ @ 2420Mhz - 1.7v: idle: 78Watt / Load: 165Watt

    In the days of the Athlon XP an increase of 0.1v vcore would result in 4-6°C higher CPU temp, so without much surprise the temperature results in this roundup with the 1.7v Athlon 64 will be much higher.

    Noise was recorded approx. 50cm away from the case at an angle, here's a (very bad) drawing of how the dBA meter was position opposite the case and the test-room.

    Madshrimps (c)
    Green box = Sonata II # White Dot = dBA meter


    What was measured?

  • The CPU temperature was measured with SpeedFan and highest value recorded
  • PWM temperature through SpeedFan, this represent the area around the CPU socket, the power management caps which you see on a motherboard, they are there to make sure the power which is fed into the motherboard coming from the PSU is filtered and delivered the CPU and other components. Too high temperature will cause Vcore fluctuations which in turn causes system instability.

    Onto the test results for the Athlon 64 Setup ->
    • next
    Comment from zerotol @ 2006/12/04
    great results from the zalman..Zalman back in the lead on great performance and silence ?
    Comment from jmke @ 2006/12/04
    it's got plenty of competition
    Comment from ilnot @ 2006/12/05
    First off, nice job, very comprehensive. I'm no fanboy of any cooler so don't misunderstand me but reading the review there is something that strikes me as odd. SO much so that I had to post.

    Upfront I'll admit that I don't know anything about any special requirements or circumstances for AMD CPU cooling but the dismal results for the Thermalright Ultra-120 are quite different from any previous review I've read. Here's just three:

    http://www.neoseeker.com/Articles/Ha...tra_120/3.html
    http://www.frostytech.com/articlevie...id=2001&page=5
    http://www.pcper.com/article.php?aid...e=expert&pid=4

    Thermalright's own XP-90 is out performing the ultra-120 in these tests, that just can't happen. From most other reviews the Tuniq and Ultra-120 are the best coolers on the market. Please, if am mistaken in some way let me know but the test results seem to indicate that there might have been some problem with the installtion.


    A kind of side question is the "Papst 120mm 4412 F/2GLL" that you use for this test, I can't find it on the Papst website. Is it the 4412FGL model. If so, the site says it operates at 55 CFM but the review says 40 CFM. Did you use a fanmate or the like to decrease the speed?
    Comment from SAMSAMHA @ 2006/12/05
    Awesome Articles
    Comment from jmke @ 2006/12/05
    I re-tested the Ultra-120 several times, the tightly packed fins + lower surface area makes it less ideal for cooling than the other tower coolers I tested in the same very low airflow case.

    a more detailed review of the Ultra-120 can be found here: http://www.madshrimps.be/gotoartik.php?articID=496



    the Papst fan I used is indeed not listed on the Papst site but it does exist, please also see SilentPCReview's review of the Papst fan : http://www.silentpcreview.com/articl...ge3.html#papst
    Comment from ilnot @ 2006/12/05
    Thank you for such a full response, I can't believe you took the trouble to retest. And also thanks for the link to the review of the fans, I am looking at a number those for my setup.The results, with lower CFM (40), makes more sense now, though I am still surprised how badly it responds to low-flo air.

    It is a great idea to use, when applicable, the same fan for all the heatsinks to get a more apples-to-apples comparison. I am curious why you use the Papst, or only use the Papst(besides it being a hell of a lot more work). As was said the in the review there are people who don't care about noise and only go for performance and those that look for the best performance at near silence. I myself (and no doubt many others) would trade off a little noise for great performance. Do you think the S-flex 1600 or this Antec http://www.us.ncix.com/products/index.php?sku=18432 or similar fan to be too loud (rated around 30 dB and 70 CFM) for most? I have never heard these fans in person. It seems like the Tuniq, Ultra-120, and any other tower-type heatsink would see quite sizable gains in performance with a minimum increase in dB levels. I recognize the desire for quiet operation but using this type of fan for these products seems to be cutting the legs out from under them. Like buying a $1200 Kentsfield and putting it with $80 RAM. An exaggeration, of course, but you get my point.
    Comment from Rutar @ 2006/12/05
    Quote:
    Originally Posted by ilnot View Post
    It is a great idea to use, when applicable, the same fan for all the heatsinks to get a more apples-to-apples comparison. I am curious why you use the Papst, or only use the Papst(besides it being a hell of a lot more work).
    its one of the better fans for noise/performance, has a good lifetime and is available at a lot of places
    Quote:
    Do you think the S-flex 1600 or this Antec http://www.us.ncix.com/products/index.php?sku=18432 or similar fan to be too loud (rated around 30 dB and 70 CFM) for most? I have never heard these fans in person.
    If you would post that into the SPCR forums you would get obliterated.

    40 CFM fans (usually around the 1200 RPM mark) are already a compromise for the real quiet freaks, they are considered quiet but not silent. Silent is around 800 RPM, where there isn't much airflow left.

    The amount of users willing to bear the noise of a Tornado or Delta has been shrinking drastically in the past 2-3 years, due to new heatsink technology and cooler running CPUs making it no longer needed to max out a chip.
    Comment from ilnot @ 2006/12/05
    @Rutar

    Are there any other fans that you would say have similar life-time reliability and noise levels?

    Also, I wasn't suggesting those fans for quiet freaks. I just think there is a segment of enthusiasts who would take 30-35 dB for performance approaching (basic) water-cooling - lazy set-it and forget-it types like me.
    Comment from jmke @ 2006/12/05
    I tested more than 20 120mm fans in 3 reviews this year
    1) http://www.madshrimps.be/gotoartik.php?articID=391
    2) http://www.madshrimps.be/gotoartik.php?articID=421
    3) http://www.madshrimps.be/gotoartik.php?articID=516

    there are plenty of alternatives to the Papst, by heart I would say Coolink, Noctua, Sharkoon, GlobalWin and Nexus, might have left out a few, check the links

    Performance wise at >1500rpm the increased airflow will make this tower heatsinks perform extremely well. do note that adding more/faster spinning case fans will have the greatest effect at the end of the day, the rear fan at 5v in my test setup is quiet but pushes little air, making it a little baking oven.

    I'm currently working on an Antec 900 Series case which features a top 200mm fan and three 120mm fans (2 front/1rear) with all Antec fans on "silent " setting the PC will be considered quiet by most users, and thermal performance is stellar.
    Comment from pyrlo @ 2006/12/05
    what about the tuniq tower compared to the new zalman cnps 9700?
    It's very difficult to compare them as they've not been on the same testbed.
    And why is the scythe ninja so much louder on the test on the last page than the test on the first page? Also the other fans on the last page seems all much louder.
    Comment from jmke @ 2006/12/05
    the reasons are all in the article, check up on the P4 test bed again

    I plan to re-test the Tuniq on the P4 setup for the next CPU roundup, in 2007
    Comment from Rutar @ 2006/12/05
    Quote:
    Originally Posted by ilnot View Post
    @Rutar

    Are there any other fans that you would say have similar life-time reliability and noise levels?

    Also, I wasn't suggesting those fans for quiet freaks. I just think there is a segment of enthusiasts who would take 30-35 dB for performance approaching (basic) water-cooling - lazy set-it and forget-it types like me.
    there are plenty of other fans, but they are hard to get and you have to choose one fan that is good for general purposes at 100%

    If you look on other general forums, you will notice that not many people are willing to put up with high fan noise anymore and there are a lot more threads about really silent fans. Especially the bad yields in additional overclock when going to a faster fan on the high end heatsinks is putting people off (which the A64 has to be blamed as it crapped out most of the time because of the chip and not the cooling)

    Core 2 scales a bit better, but not much.
    Comment from ilnot @ 2006/12/05
    Quote:
    Originally Posted by jmke View Post
    Performance wise at >1500rpm the increased airflow will make this tower heatsinks perform extremely well. do note that adding more/faster spinning case fans will have the greatest effect at the end of the day...
    Yeah, that performance is what I was getting at but I'll take your advice and look to my case fans before I go "crazy" with the CPU fan. Thanks again for all of your feedback.


    and @Rutar
    I was looking at the scalability of the C2D, this type of cooling for AMD would be pretty fruitless.
    Comment from pyrlo @ 2006/12/06
    don't forget Zerotherm coolers for the roundup in 2007. They look impressive
    Comment from FXi @ 2006/12/07
    I know end users always ask the moon, but I do have to say congrats for running inside a case. While that makes it harder to do the testing, it yields more realistic end user results. Nothing is perfect. Some will have bench setups and some will have very breezy cases, but it helps.

    I did cross reference somewhat the Water cooling article and this one. Not a perfect match but the 6400 oc'd begins to generate heat close to the old P4's at stock. Kind of a nice change, if you ask me. Anyway, seems like the Apex water cool gets a few degrees on the Zalman 9700 @ high, but not a lot more than that. In the future, it might be nice to see a toss in of not "every" cooler from past reviews, but at least a "reference" cooler or two. More and more, I think people are going to be curious where it all lands, water, air and TEC/air combo's. And using just one cooler from a previous article on the current setup, would give at least a reference point to give some rough extrapolation of the wider picture. In this case, water and a 9500 maybe? I don't want to drive you crazy with complications, so this is coming with the usual grain of salt.

    Please note secondly, that the ultra 120 with a reasonably quiet 38x120mm fan (say the Panaflo series) would be both quiet and more powerful. And the simple fact that it can use such a fan where many of the others can't fit it, that makes it quite the cooler.

    All this is food for thought. I follow all of Madshrimps articles carefully with due respect for the work involved. Have fun!
    Comment from jmke @ 2006/12/07
    Why FXi, thanks for the comments, the Swiftech review was done by a colleague reviewer who lives at the other side of the big pond, won't be easy to include in a comparison

    I always re-include a few "reference" coolers when I move to a new test setup, I rested about 5-6 on the A64 setup, and I'm re-including about the same amount on the P4 setup which each consecutive roundup
    Comment from thorgal @ 2006/12/08
    Quote:
    Originally Posted by jmke View Post
    I plan to re-test the Tuniq on the P4 setup for the next CPU roundup, in 2007
    Thank you jmke I guess my asking was not enough

    Now seriously, I love this review, it provides an in depth look at almost all (if not all) of the noteworthy cpu coolers of the last 2 years. There's always something to critisize of course, but I haven't seen many comprehensive tables around the web that even come close. Great work John
    Comment from ray_gti-r @ 2007/08/31
    I respect madshrimps and have viewed the reviews with awe for years - especially impressed by the very crisp photo's.

    BUT ... this review is junk. To use the cheapest possible Intel all-aluminium heatsink on a 20% overclocked "hot running 524" then complain that the stock heatsink/fan is "ear deafening" is complete rubbish.

    The stock Intel heatsink used in this review is only good for a stock CPU with a TDP of 69.7W.

    The 524 has a TDP of 84W. Overclocking that by 20% ... what do you expect this Intel too-low-a-spec fan to do?

    I have tested well over 200 stock Intel HSFs (all kinds, all vintages) and can confirm they are NOT "ear deafening" when used properly. My Abit IC7-MAX3 runs (UNDERCLOCKED to 2.4mhz) a 2.8c 30-capper at 41C light use @ 21c ambient using a stock Intel Extreme Edition HSF running at 1,259rpm via BIOS and only ever goes to STOCK speed at prolonged periods of 100% load. I am very happy with that. I've also tested E43xx & E63xx CPU's with matched HSf's and can confirm that those COPPER CORE HSFs are even more impressive (but NOT backward compatible with Pentium D etc CPUs - see Intel's Shop for confirmation).

    If a properly matched stock Intel Pentium 4 COPPER CORE HSF had been used to measure an oveclocked Intel CPU I'd have been more impressed with this review.

    F-

    Sorry guys!

    Cheers, Ray
    1st post - so be gentle!
    Comment from piotke @ 2007/08/31
    The stock Intel heatsink used in this review is only good for a stock CPU with a TDP of 69.7W.

    The 524 has a TDP of 84W. Overclocking that by 20% ... what do you expect this Intel too-low-a-spec fan to do?




    This cooler was provided with this CPU by intel...
    Comment from jmke @ 2007/08/31
    Quote:
    Originally Posted by ray_gti-r View Post
    If a properly matched stock Intel Pentium 4 COPPER CORE HSF had been used to measure an oveclocked Intel CPU I'd have been more impressed with this review.
    send one over and I'll test it as it stands, the 524 P4 comes with the alu HSF and that's the one I tested, don't see anything wrong, sorry buddy :/
    Comment from ray_gti-r @ 2007/09/03
    Quote:
    This cooler was provided with this CPU by intel...[/
    Quote:
    the 524 P4 comes with the alu HSF and that's the one I tested, don't see anything wrong
    Again, maximum respect to madshrimps ... as always.

    But I see two things wrong with these replies. In order of quote;-

    a) classic assumption:- the stock Intel heatsink is matched to the CPU at standard MHZ. It clearly is not. Could Intel have supplied test hardware with a mismatch problem*? Faulty Intel fan*?
    b) if you overclock, obviously you MUST upgrade the standard heatsink (DUH!). An Intel copper-core HFS is my recommendation (and maybe lap it to death & use ShinEtsu Microsi paste?).

    I also see an - um - tester problem. If the stock Intel heatsink thermal pad was not used (maybe a thin, "high performance" paste was used instead?) and the stock Intel heatsink has a poorly-finished base (rough / convex etc) then there would be inadequate thermal contact with the CPU IHS.

    I stand by what I said before ... to use a 20% overclocked "hot running 524" then complain that the stock heatsink/fan is "ear deafening" is complete rubbish. You need to check if the equipment is matched before making statements like this. Especially if you have not confirmed that the supplied heatsink is correct for the stock CPU speed ... and upped the CPU speed by 1/5th - which means _more_ than 1/5th heat generation due to thermal inefficiencies!

    *Not your fault, of course. Intel have many suppliers of heatsinks (Sanyo Denkei, Foxconn, Mitac, Delta etc) so maybe they got this one wrong, maybe the fan is defective, maybe the heatsink base is very poorly finished. But I have heard of a 524 overclocked to 4ghz using a lapped all-aluminium Intel heatsink & fan.

    Ray
    madshrips fan - when it's rite!
    Comment from jmke @ 2007/09/03
    don't quite agree with you but will add copper insert Intel HSF in next roundup
    the 524 HSF is not defective, read the review again

    "The Intel Stock cooling has a temperature regulated fan, trying to keep the CPU temperature below 60°C at all times, when we left the machine running for several HOURS at full load the fan was spinning close to 4300rpm and at 66.8dBA it’s ear deafening. If we only loaded the system for half an hour the fan would reach ~2500rpm this gave a more acceptable 50dBA noise level, on par with some of the other heatsinks tested."

    that full alu HSF is still delivered for most models and the fan used a good indicator of noise level at different temp loads, the intel stock hsf has yet to impress me, AMD has, it took heat pipes and copper core; Intel HSF, not very good noise/performance wise, not at all.
    Comment from thorgal @ 2007/09/03
    Well I actually think there are good arguments on both sides :

    - It is a fact that the Intel heatsink was not designed for heatloads like the one you're applying John. Hence the argument that it can "do nothing more but fail" is in my opinion correct. Therefor, I would not consider the Intel heasink a complete failure, but it's still too loud for my taste nonetheless.

    - On the other hand : I think the argument to say that you can't test a piece of hardware (in this case the heatsink) for a purpose for which it was not designed (overclocking), is not correct. Two reasons : in my opinion it is not necessarily bad to test a piece of hardware over their spec, to see how it behaves and to see from what point on it would fail (be that noise, temperature or whatever). Second, and most important : the stock heatsink is what you get, and many people will go to work overclocking their cpu starting with this exact HSP. It is only natural then for a reviewer to take into account the performance of this HSP, to inform the people of its capacities.

    All this just my 2c. of course
    Comment from jmke @ 2007/09/03
    Quote:
    Originally Posted by thorgal View Post
    - It is a fact that the Intel heatsink was not designed for heatloads like the one you're applying John. Hence the argument that it can "do nothing more but fail" is in my opinion correct. Therefor, I would not consider the Intel heatsink a complete failure, but it's still too loud for my taste nonetheless.
    in the performance/noise sector it certainly doesn't score very high, small fan, high RPM after extended period of stress test, sorry, it's no good.
    Comment from ray_gti-r @ 2007/09/15
    Quote:
    Originally Posted by jmke View Post
    in the performance/noise sector it certainly doesn't score very high, small fan, high RPM after extended period of stress test, sorry, it's no good.
    Make that an "out-of-spec stress test".

    Then apply it to any piece of equipment.

    That equipment will also be "no good" ... go figger.
    Comment from Sidney @ 2007/09/15
    Quote:
    Originally Posted by ray_gti-r View Post
    Make that an "out-of-spec stress test".

    Then apply it to any piece of equipment.

    That equipment will also be "no good" ... go figger.
    The tree sounds when the wind blows you got it this time.
    Comment from jmke @ 2007/09/16
    Intel stock HSF is no good. period. if you want silent computing.
    Comment from ray_gti-r @ 2007/09/17
    To quote a rather more modern gem of wisdom ... Sayng it's so doesn't make it so.

    I've got almost silent computing.
    With a stock Intel heatsink.
    Please re-read post #19 ... Since I made that post I've set my processor to stock speed (2.8ghz) and the stock Intel fan continues to run at 1,258rpm and the CPU shows 41C in Sandra Pro @ ambient 22C. My Abit 'board is reputed for overstating temperatures but that's fine with me - the Intel fan knows what's best for my CPU.

    Despite the barracking I will always stand by what I said about this review - an under-spec Intel heatsink was used on an over-spec'd CPU. Result - noise. If I've got that wrong let's agree to differ and move on.
    Comment from Sidney @ 2007/09/17
    Intel E2140 Stock 1.6Ghz default @2.93 GHz
    My 2x60mm exhaust fans in HTPC case are making more noise

    50CM away from front 40.3dBA increases to 43dBA when the stock fan tops at 2,600RPM during OCCT stress run. Not silence, and I don't need too much below 40dBA because my ambient is around 38dBA. I live in a normal house with very little traffic, and 100 ft away from my neigbors.
    Comment from jmke @ 2007/09/17
    Quote:
    Originally Posted by ray_gti-r View Post
    To quote a rather more modern gem of wisdom ... Sayng it's so doesn't make it so..
    My ears may not be top notch, but I'm sitting next to two compaq's Intel powered machines with stock Intel cooler they are noisy and it's the CPU fan, unplugging the fan drops the noise considerable. Inside a hot enviroment, read case with single or nor exhaust fan, and no in-take, the Intel HSF underperforms, as much as the AMD alu heatsink does, only if you are comfortable with high CPU temperatures and tweaking the CPU fan speed manually, can you get silence with stock cooling; in all other cases (literally and figurativly speaking) you will have a noise PC once the CPU is loaded.

    If your case features good ariflow I'm sure your system will run silent, and the fan on the Intel heatsink doesn't spin too fast to become noticeable. I've tested in extreme situation to see what the HSF is at its worse; if I were to to run the CPU at stock speeds ALL heatsinks would be DEAD silent at low fan speed without overheating; the idea of the OC was to create the heatload seen with QuadCore and DualCore CPUs, and afaik the heatsinks included the Q6600 are not too good either ( http://www.madshrimps.be/?action=get...&articID=5 96 ) .

    To end the rant, I'm going to bed, in Croatia will include the latest Intel (with copper insert) HSF in the next roundup to please the masses and my curiousity , but I'm not expecting much

     

    reply