4 pipes and a bag of fins
The design of the Hyper 48 is quite different when compared to the Hyper 6, no longer a big tower, but a smaller cube like creation with 4 heat pipes running through a set of small fins.
The whole heatsink is made out of copper with the base protruding from a solid block of copper. The layout of the heat pipes resemblances that of the XP-90 from Thermalright, with the difference in materials used and the fact that the fins on the Hyper 48 run all the way through, the heat pipes in the Hyper 48 are divided over the width of the heatsinks, while the XP-90 has them bunched up at each side. Installation
AMD K8 and Intel mounting gear
Straight forward installation was one of the benefits of the Hyper 6 and the Hyper 48 utilizes the same method for attaching to your P4 S478 or A64 S754/939/940 CPU, for P4 you can just use the default Intel bracket, but it is strongly advised to use the included back plate and separate dual platform bracket, for A64 you need to remove the default bracket and swap it with the included one, after you have done this installation is just 2 snaps away.
The Hyper 48 also adds S775 compatibility for Intel which requires you install a separate bracket at the bottom of the heatsink.Test Setup
My trusty P4 2.4 "C" was dusted off and ready for another round of torture testing:
Setup was installed in a case less environment
JMke's Test Setup
|CPU ||Intel P4 2.4 "C" @ 2.4 and 3 Ghz both with 1.55v vcore|
|Cooling ||Coolermaster Hyper 48|
|Mainboard ||Asus P4C800 |
|Memory ||1 * 256Mb PC3200 Mushkin Special Series|
|Video ||ATI R9000 Passive Cooling|
all results were taken with room temperature at ~25°C, but temp fluctuations, different mounting and user error can account up to 1-2°C of inaccuracy in the obtained results. Please keep this in mind when looking at the results. Each heatsink was tested repeatedly; if I got questionable I restarted the test.
Arctic Silver kindly send us their “Lumière” thermal testing compound which has the same color and characteristics as Ceramique, but it only needs a burn in period of 30min (compared to several days with other thermal pastes).
The CPU stressed with 2 instances of K7 CPU Burn, this application will raise temperature higher then any other stress program I know, 4-5°C difference with Prime95 or 3DMark2001SE loops.
Noise level of each HSF combo was recorded with a SmartSensor SL4001A, the sensor was placed ~60cm away. The lowest dBA reading in the test room was 33 dBA.
Onto the results ->