by William Kelley on June 23, 2010 in Air-Cooling
When Thermaltake first unveiled its FRIO CPU cooler, the company promised that it would be one of the best-performing models it’s ever released. Now that we’ve had the opportunity to put our sample through our usual battery of tests, we can attest to that claim being true. To top it off, the FRIO even includes a fan speed control and reasonable price.
To assure that our results are as accurate as possible, all of our CPU cooler testing is performed under highly-controlled conditions. Our test chassis is kept in a near-steady 20Â°C ambient environment, with readings taken before and after testing with a standard room thermometer. After we boot up our machine, we allow Windows to settle itself down for 10 minutes to stabilize processes that might be running in the background. Once the PC is completely idle, we record the current CPU temperature as that in our results.
BIOS settings are verified prior to each run, and to help with quick switching of our various profiles, we make use of the motherboard’s ability to store multiple configurations. We primarily use two for our testing here – stock speed, of 2.67GHz, and also a maximum overclock, of 4.01GHz. Stock settings were achieved by using “Load Optimum Default”, and storing those as our stock profile. The maximum overclock was obtained after extensive testing and tweaking to insure it was stable. The BCLK was raised to 191MHz. The CPU’s vCore was raised to 1.400v, and the IMC voltage was raised to 1.30v. The RAM is run at 1.6V and does overclock with the CPU during overclocking testing.
For our monitoring and temperature reporting, we use Everest Ultimate Edition 5, from Lavalys. It allows us to grab the results from each one of the cores, and the CPU has a whole, so we believe it to be indispensable to our toolkit. To help push our Intel Core i5 750 to its breaking-point, we use OCCT Linpack testing. The reason is simple: it utilizes LINPACK. After much testing with various “stress testers” in the past, we’ve found that running a multi-threaded tool that supports LINPACK, such as OCCT (and also LinX), pushes both AMD and Intel CPUs like no other. This results in higher temperatures than others (like Prime95) can muster, and also greater power consumption.
Because our test machine is equipped with 4GB of RAM, we set OCCT to use 90% of the available memory, and then set the test to run 1 hour total. With the help of Everest, the CPU’s various temperatures are recorded throughout all of the testing, and also for a minute after the test ends. The maximum recorded temperature found in the results file is labeled as “Max” in our results.
After selecting the appropriate brackets, it is a simple matter of four screws to secure said brackets to the heat sink base. Next, you install the neoprene washers and thread it through the motherboard mounting holes. Once that is done you then install the foam washers between the back plate and the motherboard. Finally the mounting nuts hold everything firmly together. I would have preferred thumbscrews instead of the included nuts which require a flat heat screw driver for proper tightening.
I am very happy to stress the fact that I was able to mount both fans onto the FRIO even though we run very tall memory on our test platform. This is a great bonus as I have had significant clearance issues with quite a few current releases and it always bothers me that I feel that overall compatibility is sacrificed by far too many designers who are more focused on absolute performance.
Space is also ample between the FRIO and the top slot of the test bed. Most motherboards come with the uppermost PCI-E slot at the top of the expansion area and this is another area where tight is not always right. Proper airflow is easily maintained and you won’t have to worry that either your graphics card or your CPU will get cooked due to clearance issues.
After the FRIO was installed and the test bed booted up, we benched these results:
Thermaltake’s results are impressive here, and are made even more so impressive since Noctua’s CPU cooler retails for about $20 more. Where the FRIO does suffer is with noise at load.