NZXT Phantom Cooling Performance

The most important task for any case is to keep your hardware cool. Whatever components it houses, regardless whether they're running at stock speed or are overclocked, if your hardware overheats, stability issues will creep in and the lifespan of all your expensive gear will be reduced. This is a particular problem if you use your PC for demanding tasks such as gaming, benchmarking or rendering, especially during the summer months.

To give the cases that we review a good workout, we updated our test kit to include an ATI Radeon HD 5870 2GB Eyefinity 6 graphics card and an Intel Core i7-870 CPU, which has been overclocked from 2.93GHz to 3.4GHz using a vcore of 1.305V and a CPU PLL of 1.9V. While this is a modest overclock, our test kit will spend most of its life under high load, so pushing it to within an inch of its life isn't in anybody's interests. More importantly, some of the poorer-performing cases topped a CPU temperature of 80°C, so even this overclock will prove to be too much for some cases.

The CPU is cooled by a Gelid Tranquillo CPU cooler with a Noctua NF-S12B ULN 120mm fan. The NF-S12B ULN is a low airflow fan, so it will highlight cases with poor cooling for the CPU. We also locked the fan speed on the graphics card to 30 per cent. The reason for this is to prevent the automatic speed control interfering with the results, making them inconsistent.

After all, we're interested in how well the cases on test are able to provide your graphics card with cool air, rather than how well the cooler on the HD 5870 2GB Eyefinity 6 graphics card works. After some experimentation, 30 per cent proved to be fast enough to keep the graphics card cool enough in poorly performing cases, while giving better cases the chance to shine.

We recorded the temperature of the CPU using RealTemp, noting the temperature of the hottest core, and used GPU-Z to log the temperature of the graphics card. Having recorded the ambient temperature in front of the main air intake of each case, we were then able to calculate the delta T, the difference between the ambient temperature of our lab and the temperature of the hardware.

As we don't test in a temperature-controlled lab, this allows for small changes in ambient air temperature, which renders absolute temperature measurements of the hardware useless. To load the system, we used a combination of the Canyon Flight benchmark in 3DMark06 and the smallfft test in Prime 95. We looped/ran each test for a minimum of 30 minutes or until the temperatures plateaued and recorded the highest temperature in the log files. The combined test represents a worst-case scenario, with both the CPU and graphics card under sustained high load.

• CPU: Intel Core i7-870 CPU overclocked to 3.4GHz
  
• Graphics Card: ATI Radeon HD 5870 2GB Eyefinity 6 Graphics Card
  
• Motherboard: Biostar TPower I55 motherboard
  
• RAM: 2 x 2GB OCZ PC3 15000 DDR3 RAM
  
• Hard Drive: 1 x 160GB Seagate Barracuda 7200.9 Hard Disk
  
• Heatsink: Gelid Tranquillo with a Noctua NF-S12B ULN fan
  
• Power Supply: Seasonic M12D SS-850EM PSU