Introduction

The product we’re testing today is not available as an off-the-shelve retail heatsink, it comes with bundled with Intel’s 45nm Extreme processors (e.g.: QX6850) . Its full name is FCLGA4-S which doesn’t make us much wiser.

Intel reference heatsinks haven’t changed much in design since the introduction of the first S478 and S775 processors. They offer a basic performance/noise level, most if not all 3rd party heatsinks offer better performance at lower noise levels so this market has been growing for several years now, we have tested over 150 of these heatsinks here at the site, and even more new CPU cooler solutions are being released at and increased rate.

When we first saw the FCLGA4-S we were quite surprised by its size and design, to say the least, it’s quite a step forward from the previous reference coolers:



The unit features a custom size 110mm fan with blue LED lightening (!) which is controlled through PWM (1000~2900rpm). When it comes to weight you might be surprised… ~420gram. If you know that the previous AlCu reference coolers weighed close to 450gram and was smaller in size you know something is up.

Our friends at FrostyTech found out what made this new cooler from Intel thick; there is a copper insert alright, but it’s hollow and serves as a heat column; which is a cheaper alternative to the well known heat pipe.

The 41mm diameter copper "thermal chamber" works along a similar principle as a heatpipe, using a working fluid under a vacuum. The low pressure causes the fluid to change states when heat is applied, allowing the vapour to rapidly conduct heat between hot and cold surfaces. A solid metal block by comparison would rely upon the metals conductive properties to move heat from the base and distribute it along the cylindrical walls to which the fins are soldered.

When AMD released their enthusiast grade reference heatsink in 2006 it was able to match some popular 3rd heatsinks in noise/performance; let us see if Intel’s fanciest heatsink can impress us ->

Closer Look

This heatsink is quite similar in design to some of the Zalman CNPS coolers we have tested in the past; the fan sits embed inside the heatsink, the aluminum fins extend towards the outer edges of the fan to increase the total surface area which can be used for cooling.

Looking at the HSF from the side reveals that the lower part is smaller in diameter to avoid issues installing the unit on motherboards which feature larger northbridge cooling solutions.



Zooming in onto one particular area of the HSF reveals a small switch which controls the fan speed; setting it to “high” will allow you the fan to speed up to maximum rpm of 2900, when you set it to “low” the fan rpm tops out at ~2000.



The transparent fan is only slightly smaller than a reference 120mm fan, which is too bad as you won’t be able to swap it out if you want to modify it with a quieter unit.



The base of the unit is copper and nicely finished. In the photo below we removed the pre-applied thermal paste which Intel puts on all their bundled heatsinks.



Installation

As you can see in the photo above, this heatsink uses push pins, motherboard removal is recommended in order to be able to reach all the push-pins.



Inside the Antec Sonata 2 the Extreme Intel HSF fits with room to spare;



Onto the test setup ->

Test Setup and Competition

We build a S775 system with parts from Alternate.de, the CPU is one hot running Pentium 4 524, 3.06Ghz. It is mounted on a Swiss-army knife equivalent of motherboards: an Asrock 775Dual-VSTA.



The mounting system on S775 is quite straight forward and well thought out, 4 holes around the socket serve as mounting points for the push pins on the standard Intel cooler. Installation is a snap, and removal is very easy too.



With the stock cooling and at stock voltage the 3Ghz P4 was running stable at 3.68Ghz, quite a nice improvement from default speeds.

A Watt Meter recorded peak power consumption under heavy CPU load at 138W, which is less than our previous Athlon 64 setup which consumed up to 165W. The Asrock bios lacks CPU voltage manipulation, so at default voltage is seems this Prescott setup is more power friendly then the over-volted AMD system.

We’re using a compact Antec Sonata II mid tower case, swapped out the PSU for a passive model from FSP rated at 400W, the outside of the PSU case never went past 40°C during our stress tests,

Intel S775 Setup
CPU	Pentium 4 524 @ 3628Mhz - 1.36v vcore
Mainboard	Asrock 775Dual-VSTA
Memory	1 * 512Mb Mushkin PC3200 LVLII V2
Other	Antec Sonata II with AcoustiFan DustPROOF 120mm @ 5v in the rear as outtake (mounted with soft-mounts) ATI R9000 Passive Cooling FSP ZEN 400W Passive Cooled PSU Seagate 7200.8 200Gb HDD in Scythe Quiet Drive

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.


example: dBA meter is placed right at the edge of the case - with side panel removed

Noise level of each HSF combo was recorded with SmartSensor SL4001A, the sensor was placed ~5cm away from the side of the case with panel removed. The lowest dBA reading in the test room was 36dBA! with system running without HSF fan.

System was stressed by running K7 CPU Burn for 30min (after Thermal Compound's burn-in); this application pushes the temperature higher than 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

Delta FFB1212VHE 120x38mm Very High Speed provided by Sidewinder Computers

151CFM – 3200RPM – 12V fan

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

GlobalWin NCB 120x120x25mm fan with 41.7CFM rating.

Delta NFB0912L 92x92x25mm fan with 42CFM rating.

Delta FFB1212VHE 120x120x38mm with 151CFM rating.

The Competition

These are the heatsinks we have tested so far on this platform and will compare the NH-U12P to:

Arctic Cooling Freezer 7 Pro

Auras CTC-868

Auras GTO-990

Auras LPT-709

Coolermaster Vortex 752

Coolermaster Sphere

Coolermaster Hyper 212

Coolink Silentator

Coolermaster Eclipse

Coolermaster Hyper TX

Coolermaster Hyper Z600

Coolermaster GeminII

Coolermaster Mars

Evercool Buffalo

Noctua NH-U12

Noctua NH-U12P

OCZ Vendetta

Rosewill RCX-Z5-Ultra

Rosewill RCX-Z775-EX

Scythe ANDY Samurai Master

Scythe Kama Cross

Scythe Katana 2

Scythe Ninja

Scythe Ninja Copper

Thermalright IFX-14

Thermalright SI-128

Thermalright Ultra-120 A

Thermalright Ultra-120 Extreme

Titan Amanda TEC

TTIC BIG

TTIC NPH-1000

Tuniq Tower 120

Ultra ChillTec Thermo Electric CPU Cooler

Xigmatek HDT-S1283

Xigmatek HDT-S1283 “Red Scorpion”

ZEROTherm BTF90

Zalman CNPS9700LED

Zalman CNPS8700

and three Intel stock heatsinks:

Intel Reference Alu (included with older Pentium 4 S775 and Intel E2xxx)

Intel Reference Alu/Cu (included with Core 2 Duo models)

Intel Reference Alu/Cu Big (included with Core 2 Quad models)

Onto the results ->

Performance Results

Here we compare the performance of Intel Extreme Reference HSF with different fan speed setting. The data in a chart showing both maximum CPU temperatures obtained under load, as well as noise levels at those settings.

We sorted the obtained data by Noise level ranges, then by lowest CPU temp. Since we measure our dBA level from very close distance, the numbers by themselves have little value without some information. Ambient noise in the room was 36dBA, so we used the following “categories”:

Extremely Noisy: >57dBA. These heatsinks can be heard from the room next door, if you are into overclocking the results obtained here will be most interesting if you don’t care about becoming deaf.

Moderately Noisy to “Not so bothersome”: All results between >44-56< dBA, while the difference between the two extremes is “twice as loud” the loudest reading at 56dBA will be do-able for most, while those who want it quieter should look at the results below 50dBA.

Quiet to Whispher Quiet: Any result under 44dBA is included here, you’ll have to focus to hear the fan running with these CPU coolers, on most you’ll be unable to do that unless you take the PC in very very quiet room and open the side panel and have no other actively cooled component inside the PC.

In the chart below we have added a few descriptions after each heatsink’s name to tell you how we tested them.

We hooked the fan up to a Zalman Fanmate2 which gives 11V at “@ High” and 5V at “@ Low”.

Some heatsinks did not allow manual fan speed regulation, those are marked with “@ Auto”

While testing some of the heatsinks with their fan "@ Low" the system overheated, instead of reporting no data, we decided to measure the fan noise and stop fiddling with the fan controller when the dBA meter read 45dBA. These results are marked “@ Low(er)” in the chart.

“+Stock” means tested with the fan included with the heatsink, “+Papst” or “+NCB” or “+Delta” or “+Delta 3200rpm” means respectively that the heatsink is tested with a 120mm Papst fan, 120mm NCB Globalwin fan, Delta 92mm or Delta 120mm High Speed fan.

We did not use the Zalman Fanmate2 with the Intel Extreme HSF, we hooked it up to 12v 3-pin connector and set the small dip-switch to “high” and “low” settings to obtain our results.



Results explained and conclusive thoughts on the next page ->

DATA Analysis

Performance wise the new Intel Extreme Reference HSF was the best of all Intel HSF tested, improving temperatures by as much as 8°C. At this performance level the heatsink ranks on par with popular 3rd party solutions as the Scythe Ninja and OCZ Vendetta. When we put the fan speed to “Low” the load temperatures increased by ~4.5°C.

So looking purely at the temperature results this heatsinks does quite okay; but when we include the noise readings this outcome is less impressive. To get 53°C load temperatures the fan is making quite a bit of noise, ~64dBA (at 5cm) and this puts it into the highest noise bracket of the chart; If you realize that the Scythe Ninja gets 53°C at 47.3dBA you understand that the Intel cooler is very noise indeed.

Switching the fan speed to “Low” the noise dropped to a merely “annoying” 56.3dBA level… the list of 3rd party heatsinks which offer better performance at lower noise levels than this is very long.


Blue LED lightening doesn’t make it less noisy unfortunately

Conclusive Thoughts

Overall we weren’t quite impressed with the performance numbers of the Intel Extreme Reference Heatsink (what a mouthful!). While pure temperature results were commendable for its compact size and low weight, the fan makes too much noise to make it a viable solution for “acceptable” CPU cooling.

If you don’t care about noise and you HSF weight and size is critical for your choice, you could try to find this unit on eBAY or similar service. If you value quiet CPU cooling with acceptable performance you should look elsewhere as this Intel cooler doesn’t improve on the performance/noise balance you get from many affordable 3rd party heatsinks.

+ Free with 45nm Extreme Processors
+ PWM fan control
+ Low weight

- Very noise fan
- Below average performance

We thank Thorgal, our local memory module reviewer, who allowed us to test his reference heatsink.