When enthusiasts pick a new CPU or video card, one of the main things they look at is how the part performs and what sort of price/performance ratio it has. Benchmarks are typically used to quantify these metrics, but can the hard data spit out by benchmarks be trusted? It's a question that pops up from time to time. In the past we've had plenty of stories of manufacturers cheating in benchmarks, or benchmarks that showed a certain bias. But other than this, a lot also depends on the settings or quality of the benchmark, as it doesn't always reflect real in-game conditions.
In a new video, DigitalFoundry takes a look at the current state of affairs, and ponders over whether typical in-game benchmarks are useful.