Testbed and Methods
All tests were performed inside a system case with the side panel removed. Our testbed was identical for all coolers throughout the test session and featured the following configuration:
- Mainboard: ASUS P6T Deluxe (Intel X58 Express), LGA 1366, BIOS 1701;
- Processor: Intel Core i7-920, 2.67 GHz, 1.25V, 4 x 256 KB L2, 8MB L3 (Bloomfield, C0);
- Thermal interface: Tuniq TX-2;
- Graphics card: ZOTAC GeForce GTX 260 AMP2! Edition 896 MB, 648/1404/2108 MHz (1030 RPM);
- Memory: DDR3 PC3-12800 3 x 2 GB OCZ Platinum Low-Voltage Triple Channel (Spec: 1600MHz / 7-7-7-24 / 1.65 V);
- System HDD: Western Digital VelociRaptor (SATA-II, 300 GB storage capacity, 10,000 RPM, 16 MB cache, NCQ);
- Backup HDD: Western Digital Caviar Green WD10EADS (SATA-II, 1000 GB, 5400 RPM, 32 MB, NCQ);
- Scythe Quiet Drive 3.5” HDD silencer and cooler chassis;
- Optical drive: Samsung SH-S183L;
- System case: Antec Twelve Hundred (front panel: two Noiseblocker NB-Multiframe S-Series MF12-S1 fans at 900 RPM and Scythe Gentle Typhoon fan at 900 RPM; back panel: one Scythe Slip Stream 120 fan at 900 RPM; top panel: standard 200 mm fan at 400 RPM);
- Control and monitoring panel: Zalman ZM-MFC2;
- Power supply: Zalman ZM1000-HP 1000 W (with a default 140 mm fan).
During this test session we managed to overclock our 45nm quad-core processor with the leveled out heat-spreader surface using 21x multiplier and enabled “Load-Line Calibration” to 3.95 GHz using the weakest cooling system of the today’s testing participants with the fan working in quiet mode. The nominal processor Vcore was increased to 1.35 V in the mainboard BIOS.
The memory voltage was at 1.62 V and its frequency was around 1500 MHz (7-7-7-14_1T timings). All other parameters available in the mainboard BIOS and connected with CPU or memory overclocking remained unchanged (set to Auto).
All tests were performed under Microsoft Windows 7 Ultimate RTM x64. We used the following software during our test session:
- Real Temp 3.30 RC11 – to monitor the processor core temperature;
- Linpack 64-bit with LinX shell version 0.6.3 – to create maximum CPU load (two test cycles, 5 Linpack runs in each cycle with 3584 MB RAM capacity involved);
- RivaTuner 2.24 – to visually control temperature changes (with RTCore plugin).
- CPU-Z 1.52.2 – to monitor processor core voltage and frequency.
So, the complete screenshot during the test session looks as follows:
The workload was created with two consecutive Linpack runs with the settings described above. The additional tests showed that increase in the memory capacity used by Linpack x64 causes minimal increase in the peak CPU temperature, which occurs only at low fan rotation speed of the liquid-cooling system (800 RPM). In other words, we chose sufficient amount of memory for our tests with Linpack x64 as well as sufficient number of iterations. You can see it clearly from the screenshot taken right after we completed the test, namely the RivaTuner graph with automatically marked maximum processor temperature. We reach peak temperature on the second iteration already and the next three iterations didn’t heat up the CPU any more.
The stabilization period for the CPU temperature between the two test cycles was about 8-10 minutes. We took the maximum temperature of the hottest processor core of the four for the results charts. Moreover, we will also provide the temperatures of all cores in a separate table alongside with the calculated average value. The ambient temperature was checked next to the system case with an electronic thermometer with 0.1 °C precision that allows monitoring the temperature changes over the past 6 hours. During our test session room temperature varied between 20.8-21.1 °C. When we performed the maximum CPU overclocking tests with all participating liquid-cooling systems, the room heated up to 23.0 °C (the heater was on because of the cold weather outside).
The noise level of each cooler was measured after 1:00 AM in a closed room about 20 m2 big using CENTER-321 electronic noise meter. The noise level for each cooler was tested outside the system case when the only noise sources in the lab were the cooler and its fan(s). The noise meter was installed on a tripod and was always at a 200 mm distance from the cooler. To measure the noise we set the cooler onto a 45 mm stand made of polyurethane foam material on top of a desk. The lowest noise reading our noise meter device can register is 29.8 dBA and the subjectively comfortable noise level in these testing conditions was around 34.5-35 dBA. The fan(s) rotation speed was measured in the entire supported range using our controller by changing the voltage.
Corsair H50 will be competing against our reference super-cooler - Thermalright IFX-14, which base I have finally evened out and polished:
The cooler was equipped with two Thermalright TR-FDB-2000 fans working in moderate acoustic mode at 1000 RPM (±10 RPM) and at maximum rotation speed of 2000 RPM. The fan rotation speed was adjusted and monitored using the above mentioned controller. Corsair H50 was tested with the default fan as well as with two Thermalright fans, too.