I just built a Doom 3 gaming PC with two 36GB Western Digital Raptor drives set up in a RAID 0 array. Seven days ago, I would have sworn up and down that because of RAID, the level load times for id Software's latest version of Doom were significantly faster than they were on a single-drive system of similar configuration. In fact, I found myself thinking, Man, it's amazing how big a difference a RAID array makes.
It turns out that I, like many, many other gamers and PC manufacturers, was wrong. Level load times in Doom 3--along with scores of other modern PC games--are no faster with RAID than with a single drive. In fact, in some cases, level load times are actually slower with RAID than with a single drive! ...
Associate Editor Josh Norem literally stumbled upon an Internet story on a hardware site that detailed a finding in direct contrast to our long-held belief that RAID 0 provides faster read times, and therefore faster level loading times. If you're a hardcore PC user, it's quite likely you hold this belief yourself.
Norem and Senior Editor Gordon Mah Ung immediately set out to verify the results in a frenetic three-day period of testing, speculation, and bickering. Adhering to the principles of scientific method, which dictates thorough and repeated experimentation until one has validated or proved wrong the hypothesis--in this case, that RAID 0 is no faster than a single drive--the editors tested multiple system configurations with a slew of different games. The results of their testing supported the reports we originally read online.
(From October 2004 Maximum PC, written by Editor in Chief George Jones.)