You are here

CoolerMaster Hyper 6 Heatsink for P4/K8

June 9, 2004 by Edward Ng with Mike Chin

Product
Hyper 6 (KHC-V81) P4/K8 heatsink
Manufacturer
Cooler Master
Market Price
~US$55

Cooler Master is a company that needs no introduction to computing enthusiasts. Known primarily for richly-featured aluminum enclosures, Cooler Master has long been a producer of CPU coolers. Performance and style have always been important for Cooler Master; low noise has not been a prominent objective with their products in the past. This helps to explain why SPCR has not reviewed a Cooler Master heatsink before.

The Cooler Master Hyper 6 is a new cross-platform (P4/K8) heatsink promoted as a "high performance and ultra silence" product. Because of its large size and sophistication, the Hyper 6 looks like a good candidate for high cooling performance with a quiet, low-airflow fan.

Hyper 6 is a tall, large, copper heastsink. It makes use of heatpipes, which Cooler Master began incorporating in some of its products last year. The Hyper 6 bears a rough structural similarity to the Heatlane Zen NCU-1000 cooler, but the latter is mostly aluminum and does not have fan mounts, as it is meant to be run fanless. The Hyper 6 is actually one of a more recent wave of tall, large, copper / heatpipe heatsinks with fins parallel to the motherboard, and the fan mounted perpendicular to it. These include:

  • Aerocool Deep Impact DP-102
  • Gigabyte 3DCooler-Ultra
  • Thermaltake Silent Tower
  • Thermal Transtech TTIC NPH-2 Heatpipe Heatsink
  • Auras T6C Pentium 4 Heatsink
COOLER MASTER HYPER 6 SPECIFICATIONS
Socket Type
AMD K8 (socket 754/940) and Intel P4 (socket 478)
Heat Sink Dimensions
96x82x120 mm
Heat Sink Material
6 heat pipes + 100% Copper stacked fin with copper base
Fan Dimension
80x80x25 mm
Weight
~1 kg; fan adds ~100 grams
Fan Speed
1800 ~ 3000 rpm
Fan Life Expectance
40,000 hrs
Bearing Type
Sleeve bearing
Voltage Rating
6 ~ 12V
Noise Level
21 ~ 34 dB(A)
Connector
4 Pin (Power Input), 3 Pin (Speed Detection)
Application
P4 and K8

DESIGN DETAILS



Hyper 6 shown with complete mounting system.

See the four screwholes for the fan?

There's another set of fan mount holes on the other side.

Hyper 6 is a complex design:

The bottom portion is composed of a thick copper base and what appears to be almost a complete conventional heatsink bonded to it with two machine screws, and perhaps soldered as well. This sub-heatsink (for want of a better term!) appears to be made of copper-coated aluminum and has 11 thick, widely spaced
fins.

Six L-shaped heatpipes are embedded between the base and the sub-heatsink. The drawing below shows the arrangement of these heatpipes.



Heatpipes soldered
and clamped to base or only clamped?



Little to complain about the polished copper base: It is flat and smooth.

The upper stage consists of 27 thin (approx. 0.2mm) 96 x 82mm copper
fins spaced about 1.75mm apart from each other, and stacked parallel to the base. Each of these fins is bonded to the six heatpipes. The heat reaches the fins by way of
vapor-phase change occurring within the six copper heatpipes as a result of the CPU heat under the base.

There are other parts:

An aluminum frame is wrapped around the sides and top, and provides mounting points for the fan, which blows parallel to the base. There are fan mounting holes on both of the open sides, though only one fan is provided. This means push-pull, dual-fan cooling can be easily set up.


Aluminum frame fitting at side / bottom of base (left) and on the top (right).

The mounting system uses a sturdy backplate that goes under the mainboard, a fiber-based polymer frame that mounts
on top of the mainboard by screwing into the backplate, and finally,
two strong metal clips that hold the heatsink down to the fiberglass frame.
These clips are used whether the heatsink is mounted on a PC4 or a K8 board. This hardware is compatible with either P4 or K8 platforms. The top retention frame fits on either P4 or K8 boards. The backplate goes on one way for P4 boards, and is neatly flipped over to the other side for use with K8 boards.



The fitting work with both P4 and Athlon 64 (K8) CPUs.

The mounting system has to secure the >1 kg weight of the heatsink and fan. This is more than double the recommended maximum for either AMD or Intel CPU boards.

The mounting system survived multiple installations and un-installations during testing. Whether the long term stress of the high mass on a vertical mainboard in a tower case would damage the board is difficult to say. The cantilever force of this much mass is undoubtedly high. You don't need a catastrophe for damage to occur; undetectable hairline fractures are enough to make multi-layer printed circuit board unusable. Great care is advisable when handling a mainboard with this much mass hanging off it. It's probably better to use it in a horizontal desktop rather than a vertical tower.

The Fan is a 80mm, sleeve-bearing 80mm diameter unit made of translucent plastic.
An inline 4-pin Molex power connector runs out of it, as well as a single-wire 3-pin
hook-up for speed monitoring, and finally a small speed dial with a notch
on the side. This dial can be retrofitted to either a PCI slot backplate, or
a brushed aluminum 3.5" bay cover, both of which are included with the
package.

The stock fan has special 1/4" bevels cut into the intake side of the
frame, and the four short screws included are long enough to mount the fan only when
it is oriented to blow into the heatsink. A visit to Home Depot revealed that the screws have a much finer thread than
any they carry. More on the significance of this later.

How Heatpipes Work

Inside a heatpipe is a liquid that boils into vapor when it gets hot. This vapor releases the heat at the cooler surfaces of the heatpipe then condenses back into liquid. The liquid flows back down to the hot end by capillary action and gravity. Heatpipes are capable of transferring a large amount of heat per volume of working fluid due to the phase change (liquid-to-gas-to-liquid-ad infinitum) that takes place. In layman's terms, it's vaguely like watercooling without the pump, but better. Here is Cooler Master's explanatory diagram:

With the Hyper 6, the only configuration in which
the heatpipes don't work at their best when it is
mounted upside down — which is highly unlikely. Every other angle or position is fine.

INSTALLATION

Mounting the HS requires that you have access to both sides of the motherboard. This usually means the motherboard must be not installed in a case, although some hardcore enthusiasts have boasted of cutting a hole on the motheboard tray of the case to gain acess to the back side of the board without removing it.

For this review, an ASUStek P4P800 P4 board was left out of the case. The stock
heatsink retention frame was carefully removed.
The backplate was placed underneath the board and properly aligned to the four motherboard holes. Then the dedicated HS retention frame was fastened with four supplied screws with the mainboard sandwiched between retention frame and backplate. The photo below shows the dedicated HS retention frame and the CPU with Arctic Silver V thermal interface material (TIM) applied.

The dedicated retention frame, you will notice, has square holes in four corner posts that look very similar to the standard P4 frame. These corner posts are used whether the Hyper 6 is mounted on a P4 board or a K8 board. The clip is more fiddly to install than it appears. You do need two hands. Once on, it seems secure.

The photo below shows the Hyper 6 heatsink in place, with clips fastened on either side. The color has been bled articificially from the photo except around the clip so its details can be better seen.

While the retention frame appears to be stronger than the standard one on most P4 motherboards, the reliance on those plastic corner posts to support all that weight seems more about engineering convenience than optimal design. With a K8 board, only two bolts secure the retention frame to the motherboard, which is a bit less secure. Still, the clips and retention frame survived multiple installations and un-installations during testing.



Hyper 6 completely mounted and ready for testing.

TESTING

1) The Test Platform

The test was conducted in an air-conditioned room. The test
bed consists of components from my digital imaging / desktop
layout workstation, Alpha
Three
:

  • ASUStek P4P800 P4 mainboard
    with stock passive northbridge cooler
  • Intel Pentium 4-2.6C at stock voltage and speed
  • 2x 512MB PNY Verto PC3500 DDR SDRAM
  • Matrox Millennium G550 passively cooled graphics adapter
  • 160GB Seagate Barracuda 7200.7 HDD attached to a Zalman
    ZM-2HC1
    and placed on soft rubber isolation grommets
  • Seasonic Super Tornado 350 PSU

The mainboard was placed on top of an antistatic
bag on a metal table, alongside the hard drive. The
onboard LAN, USB and SATA controllers were also disabled during the test, and
a barebones Windows XP build with most services disabled was installed on the drive
in order to perform testing. The Antec Sonata case was flipped upside
down and placed in front of the setup so that the power switch could still
be utilized.

2) Instrumentation

  • An Extech
    22-816
    True RMS Multimeter & Temperature Probe
    was utilized during the
    test; the thermal probe was positioned about a foot to the side of the motherboard to monitor ambient temps.
  • Motherboard Monitor 5 (MBM5) was used to
    log temperatures during the test.
  • Zalman Fanmate1 fan controllers were used in conjunction with the Extech multimeter to set fan voltages.

3) Test Procedure

As established in previous SPCR heatsinks reviews and in SPCR's Unique Heatsink Testing Methodology article, a Panaflo FBA08A12L1A 80mm low speed fan was used. The basic method is to conduct all tests (whenever possible) with this quiet fan at 12V, 7V and 5V to eliminate performance differences due to higher or lower airflow fans. By eliminating the fan as a variable, we measure the cooling power of the heatsink itself, especially with quiet, low airflow.

One problem with using the reference Panaflo fan is those odd screws mentioned earlier. An alternative was to use some small plastic zap straps (plastic
cable ties). Although
it took quite a bit of pushing and pulling with needlenose pliers, I managed to get the Panaflo securely mounted for the test.

Given the ease of mounting two fans, the Hyper 6 was also tested with two Panaflo 80L fans, one blowing in on one side and the other blowing out on the other side. The pressure through the fins is increased, which increases effective airflow for better cooling. Few heatsinks make push-pull fans easy to implement, so we had to give this a try.

The Hyper 6 was also tested with the stock fan as a check of Cooler Master's claim of ultra silence. Checking with the multimeter, I found the stock Cooler Master fan speed controller provides
a voltage range of just 12V~10V. No wonder the control hardly seemed to change fan speed or noise! I left the controller on High and used a 3-pin to Molex converter in conjunction with the Zalman FanMate1 to control the thing.

A Thermalright SP-94 copper heatsink, already tested in a previous review on a different test plaform, was tested as a reference to compare results, and as a check of the test platform.

3) Other Details

  • Ambient room temperature during testing was 24°C.
  • Motherboard temperature sensor error: The ASUStek P4P800 P4 board is known to underreport the CPU diode temperature by ~8°C. All the temperatures in the tables below are compensated by adding 8°C to the readings reported by MBM5.
  • Idle temperatures were recorded after 30 minutes of idling in Windows.
  • Load temperatures were recorded after stressing the CPU for 30 minute with CPUBurn.
  • Motherboard Monitor 5's log function was used to record temperatures..

4) Key Results

The most important results of any heatsink tesing are:

A) The temperature rise over ambient allows us to examine the raw temperature data from different tests with the same CPU. It is very difficult for the test environment temperature to be held constant at all times.

B) °C/W — °C rise over ambient per watt of CPU heat — is calculated by dividing the temperature rise over ambient by the heat (in watts) of the CPU. The lower this number, the greater the cooling power of the heatsink. It can be used to predict the maximum temperature with CPUs other than the one used to test the heatsink. It's also allows a fair comparison between heatsinks, regardless of ambient temperature or CPU, as long as the same fan/speed is used.

The number used for W, the power dissipated by the CPU, has a major effect on °C/W. For P4 heatsinks in SPCR reviews to date, we have used Intel's Thermal Design Power (TDP) specification. Questions about the P4's actual maximum power dissipation have been floating around ever since Intel first devised the TDP.

Processor Electrical Specifications, our favorite CPU power dissipation reference, quotes Intel on TDP as being the

"worst case power dissipated by the processor while executing publically available software under normal operating conditions at nominal voltages that meet the load line specifications... The Intel TDP specification is a recommanded design point and is not representative of the absolute maximum power the processor may dissipate under worst case conditions... Processor power dissipation simulations indicate a maximum application power in the range of 75% of the maximum power for a given frequency."

We recently found another website, CPUHeat & CPUMSR Projects, which has calculated the Maximum Power Dissipation of Intel processors:

"Intel hides real power consumption behind Thermal Design Power. TDP is a power consumption of a processor while executing normal software. That is not while executing a stress test software like BurnK7.

"While TDP may be a useful number for CPU cooler manufacturers, it's not useful for end-users. This is because in real world use, there can be application that forces Intel processor to drain more power than TDP."

They have compiled is a list of computed maximum thermal power of Intel processors based on TDP number.

We believe CPUHeat & CPUMSR Projects' Maximum Power dissipation (MP) is a more accurate estimation of P4 CPU power dissipated during SPCR heatsink testing. We will be transitioning to use MPD figures to calculate °C/W numbers for P4 heatsinks. To ensure the data from previous HS reviews can be compared to new ones, two sets of °C/W figures will be presented, one based on the TDP for the P4-2.6C (69W) and one on the more realistic MP (78W).

TEST RESULTS

First with the stock fan...

Hyper 6 w/stock fan
Fan voltage
°C Idle
°C Load
°C Rise >ambient
°C/W re TDP
°C/W re MP
12 V
32
40
16
0.23
0.21
7 V
33
43
19
0.27
0.24
5 V
35
47
22
0.32
0.28

The performance at all fan voltage levels is excellent, but the noise is terrible. At 12V, the stock fan exhibits a low-mid frequency
motor noise that is audible anywhere in the test room, and even
a decent distance right down the hall when the door is open. Worse than those
acoustics, however, is the vibration; this fan shakes like it caught the bad
episode of Pokemon and went into convulsions. Attached to the Hyper 6, the vibration conducts into the entire
assembly, further exacerbating noise by exciting the entire testbed
into a full sound and fury fiesta.

Absolutely, positively no SPCR community
member in their right mind would ever want this thing blasting away at full
throttle in their rig. The thing that really surprises me is Cooler Master's
decision to not utilize one of their Rifle Bearing fans, which,
from my experience with two samples so far, is a dramatically superior fan.

While much better behaved at 7V, the stock fan is still
not quiet. Vibration from the stock fan diminishes, but
it is still there. Even at 5V, while vibration is finally insignificant, the motor noise is still plainly audible. So much for ultra silence.

With the Panaflo 80L...

Hyper 6 w/Panaflo 80L fan
Fan voltage
°C Idle
°C Load
°C Rise >ambient
°C/W re TDP
°C/W re MP
12 V
32
41
17
0.25
0.22
7 V
33
46
22
0.32
0.28
5 V
35
55
31
0.45
0.40

What blessed relief after the stock fan! At 12V, the Panaflo almost matches the performance of the stock fan at subjectively half the noise. But for most SPCR readers, it is still too noisy. At 7V, the cooling performance is still excellent, and the noise drops to a level that many would judge as very quiet. You can see the dramatic increase in temperature as the airflow drops to

With dual Panaflos 80L in Push-Pull...

Hyper 6 w/Panaflo 80L x 2 fans
Fan voltage
°C Idle
°C Load
°C Rise >ambient
°C/W re TDP
°C/W re MP
12 V
32
39
15
0.22
0.19
7 V
33
44
20
0.29
0.26
5 V
35
52
28
0.41
0.36

With one fan blowing in on one side and another fan blowing out on the other side, the pressure through the fins is increased for better cooling. The noise penalty is theoretically +3dBA. Subjectively, it is a small increase. Except at 5V, this configuration essentially matches the excellent cooling of the stock fan but at much reduced noise levels.

Is there enough room for a second fan?

The answer depends on the location and juxtiposition of the CPU HS retension bracket on the motherboard. If the heatsink retention bracket's long side is parallel to the I/O panel of the board, then there will not be enough room: The second fan will probably extend beyond the edge of the motherboard and jam up against the power supply. This is precisely the situation with my ASUS P4P800 mainboard; check the photos in previous pages.

If it is rotated 90° so that the short side of the HS retention bracket is parallel to the I/O panel of the mainboard, then there probably would be enough room for the second fan in most cases. This configuration of the mainboard is definitely preferable for the Hyper 6. See next text box for details.

COMPARISONS

First, here are the results for a Thermalright SP94, highly ranked in previous SPCR reviews.

Thermalright SP94 w/Panaflo 80L fan
Fan voltage
°C Idle
°C Load
°C Rise >ambient
°C/W re TDP
°C/W re MP
12 V
32
45
21
0.30
0.27
7 V
33
53
29
0.42
0.37
5 V
35
73
49
0.71
0.62

And now, a comparison between the Hyper 6 and the Thermalright SP94 both equipped with the Panaflo 80L.

Hyper 6 versus SP94 w/ Panaflo 80L
Fan Voltage
Heatsink
°C Idle
°C Load
°C Rise
°C/W re TDP
°C/W re MP
12V
Hyper 6
32
41
17
0.25
0.22
SP94
32
45
21
0.30
0.27
7V
Hyper 6
33
46
22
0.32
0.28
SP94
33
53
29
0.42
0.37
5V
Hyper 6
35
55
31
0.45
0.40
SP94
35
73
49
0.71
0.62

It's a downright massacre, isn't it? The cooling advantage of the Hyper 6 is 4°C at 12V, which is pretty big as heatsink performance goes. But it gets bigger as airflow is lowered: 7°C at 7V and a whopping 18°C at 5V.

In fairness to the SP94, previous experience and testing have shown that it does much better with a 92mm fan, for which the design appears to have been optimized. So, let's do another comparison, this time Hyper 6 + Panaflo 80L vs. SP94 + Panaflo 92L. The Panaflo 92L is rated for 43 CFM while the Panaflo 80L is rated for 24 CFM (at 12V); the manufacturer's noise ratings are 27 and 21 dBA/1m.

Hyper 6 + Panaflo 80L VS. SP94 + Panaflo 92L
Fan Voltage
Heatsink
°C Idle
°C Load
°C Rise
°C/W re TDP
°C/W re MP
12V
Hyper 6
32
41
17
0.25
0.22
SP94
30
41
17
0.25
0.22
7V
Hyper 6
33
46
22
0.32
0.28
SP94
31
46
22
0.32
0.28
5V
Hyper 6
35
55
31
0.45
0.40
SP94
33
55
31
0.45
0.40

Well, we have a tie in cooling performance. The Thermalright SP94 is able to match the performance of the Hyper 6, but with a >50% airflow advantage and higher noise. It's true that the noise difference at 7V and 5V is much smaller than the 6 dBA difference at 12V, but it is still there; the 80L is a quieter fan.

ANALYSIS and CONCLUSIONS

Cooler Master has achieved a new level of performance for an air-cooled heatsink. On the test bench, the Hyper 6's margin of superiority over the excellent Thermalright SP94 is nothing short of amazing, especially with lower, quieter airflow, which is of greatest interest to SPCR readers.

Victory in the real world, however, is not so concrete.
Let's look at the Good, the Bad and the downright Ugly about Hyper 6.

The Good: Hyper 6 is a capable cooler, no doubt about that, particularly when
things really heat up. In the end, a heatsink's job is
to take heat away, and Hyper 6 definitely serves its purpose well!

The Bad:

1. The stock fan is a joke, it is far too loud. Surely, Cooler Master must reconsider their choice and include a quiet rifle bearing fan instead.

2. It is a bad idea to restrict users
to the stock fan.
Why not provide some extra hardware to make using other fans (including a second fan) easier?

3. The fixed "directionality" of the fan will be a problem with many motherboards. On the ASUS P4P800 mainboard, for example, the fan ends up blowing hot air towards the power supply regardless of whether it is installed in a desktop or tower ATX case. This means the hot air off the CPU/heatsink will envelope the power supply, cause the PSU to heat up, and probably ramp up the PSU fan. A typical exhaust fan on the back panel will have to work hard (probably at higher speed) to catch any of that heat and blow it out the back.

It is only if the HS retention bracket is rotated 90° (from the configuiration on the ASUS P4P800) that the direction of the fan airflow makes sense. A back panel exhaust fan can then easily capture the heated air and blow it out of the case, resulting in that effective push-pull dual-fan cooling configuration for the heatsink. [Editor's Note: It's hard to say how many P4 boards have the preferred HS retention bracket configuration for the Hyper 6. The four P4 boards in my lab all have the wrong configuration, like Ed's Asus P4P800.]

Most Athlon 64 boards do seem to have the HS retention bracket oriented in the right way (for the Hyper 6). The location of the CPU nearer the middle in most A64 boards also makes mounting two fans on the Hyper 6 or running an exhaust duct to the back more feasible.

A simple fix is to allow the user the option to reinstall the aluminum shroud turned 90° so that the fan can blow towards the back panel regardless of retention bracket alignment. It would require square fins; they may not be square now.

The Ugly:

A whole kilogram?!? Does anyone even want to imagine what could
happen if this thing broke loose? It would almost certainly take out
that shpankity $500 Radeon X800 XT PE you just installed to go along with your
new bad boy overclocked P4!

As if the weight isn't bad enough, the tall height of the heatsink acts as a cantilever to apply even greater torque to a vertically mounted mainboard. Tower cases are still favored by the power user market this heatsink addresses. Can you trust the
HS mounting system over the long term? Can long-term stress on the motherboard cause a microscopic break in the mutili-layer PCB? Will the mounting points between the motherboard and case hold? These questions cannot be answered easily.

One thing is for sure: Remove the Hyper 6 before transporting the system to
prevent damage to the mainboard and other components.

Cooler Master should consider a lighter version with thin high grade aluminum fins. It would probably provide 98-99% of the cooling power but drop the weight by at least 40%.

The Hyper 6 is an outstandingly powerful
P4 and K8 cooler that relies on state of the art design and
brute size. With a suggested retail price of US$55, this heatsink is actually
quite a good value for those with a super hot processor and a motherboard with the right retention bracket alignment. Because of its weight, however, it is best used in a desktop
form factor system, and we suggest you heed seriously the cautions in this review.

Many thanks to Cooler Master for providing
this Hyper 6 review sample!

* * *

Discuss this review in the SPCR Forum.

Sections: 

Google

www SPCR