Anytime you use new computer components, you are subject to the effects of manufacturing defects, shipping damage, and many other possible sources of component failure. A "burn in" process stresses these components, so that if they are close to failure, we can make sure they break here so we can fix or replace them. It is much like the way Boeing puts huge stresses on aircraft wings before flying the aircraft -- if the wings have a defect and are going to fail, they want that failure to happen before the aircraft leaves the plant. So, a burn in is a way of ensuring that only hardware without defect leaves our facility.
What exactly do we do? The first tests we run are very targeted, and put a high amount of stress on individual components. Memory is tested with the utility from MemTest86.com, and the CPU is tested at the same time (MemTest86 also stresses the CPU). The hard drive is tested with the manufacturer's factory certification utility. This all takes about 6 hours. Windows is then installed, and all driver updates are installed. At that time, we run dozens of benchmarks and games, to provide benchmark data that we record and use to make sure your PC is performing as expected. We use two utilities with the specific intention to push the hardware to its limits -- FurMark and Prime95. FurMark stresses the video card to an extreme degree (so much that we do not recommend running it outside a controlled environment), while Prime95 tests the CPU cores and memory. These utilities are often left looping when the system is in standby for the next step in our build process. Running these utilities at the same time stresses all aspects of your system simultaneously, and causes the system to run as hot as it can possibly run. We monitor the temperature of your system to ensure adequate cooling, and to ensure stable operation. Most problems uncovered in our burn in process are found very early -- any memory problems are usually found within 5 minutes, and any temperature problems within an hour.