How did I test this little unit ? First I wanted to test it out of the box. No extra fans or such added, this to simulate the average end user experience. Not everybody has the luxury to have front, top and side panel coolers. The 2nd test more airflow was generated via a top fan sucking air out of the case. Thirdly to improve the unit performance I added a second fan for a push and pull configuration. Note that I still use the original fan together with an Antec Tricool fan at medium setting (1600RPM). The 2nd Push and pull configuration comprises of 2 Antec fans, both running at full blast (2000RPM). Final test with MX2 instead of the stock TIM and the Antec fans mounted on the radiator.
Testing program is good old prime95. I select custom test, with 20K fixed. The rigs are allowed to run idle for about 15 minutes before being primed to death during an hour. With the previous Core i7 water block roundup
we took a 2 hour prime test, but hottest temps were easily reached within the first hour. So one hour of prime is more than sufficient to get a good idea how the H50 will perform. Many forum members asked me why not to use LinX. I'm well aware that LinX or IBT stresses the CPU more than any other program out there, so it generates more heat then Prime95, at the same time Prime95 will resemble real application load temperatures.
Competition for the H50 will be in form of the Thermalright TRUE equipped with a 1600rpm 120mm fan and the stock Intel cooling.
First up, the E7400 running at 4ghz ( 10 x 400FSB ) 1.35Vcore
Out of the box the H50 unit is on par or just beating the Thermalright eXtreme cooler. The boxed cooler is trailing here by almost 20°C. Installing the top 120mm fan in the case didn't help much. I suspect the OCZ power supply does already a good job of getting hot air out of the Antec P160 case. Once I installed a 2nd fan on the radiator (1600rpm setting) I gained another 1°C. Boosting the fans speed to 2000rpm didn't help at all. The idle temps got a degree or 2 lower, load temps stayed the same. Overall Corsair did a good job of selecting a 1600rpm fan , combining enough airflow with acceptably silent operation. Finally replacing the TIM with MX2 gave me another small gain. I reseated the latter 3 times to ensure proper contact.
I7 up next, 3.8Ghz running on 8 cores at 1.15Vcore
A good cooler does miracles on an I7 CPU. These are monsters, wattage to heat converters
we call them. The stock cooler is almost making the 950 throttle and the test was abandoned after 10 minutes. It just couldn't handle the heat. The TRUE and the stock H50 unit perform at least 25°C better under full load. When trying to improve the H50's performance we see the same results as with the S775 setup. Adding a second fan is the best and easiest thing you can do.
Here's a screenshot with the H50 handling the Core i7 firstname.lastname@example.orgGhz quite well during some Single and SLI 9800GT runs...
(click to zoom)