OT: Heat Kills
Disclosure: I'm short AMD, and I noticed this long-winded message reflected this in the last long paragraph!
I work in the silicon reliability area, and can echo the statements of the problem of heat, as well as AMD's particular problem with it. Tom's Hardware has a great story on it, including what happens when one removes the heatsink on versions of AMD and Intel chips (summary: AMD chips are actually destroyed; Intel handles it). Don't miss the video--it' amazing to watch!
www4.tomshardware.com
Most failure mechanisms are thermally accelerated, which is to say, the hotter the chip runs, the sooner it eventually fails (and this is non-reversable). Indeed, to simulate ten years of stress in short about of time, most companies run the chips in an oven at elevated voltages--otherwise, we'd have to wait 10 years to see if they still work.
Anyway, with that in mind, I suspect there is a correlation between "AMD probably makes the faster processors on Earth" and "AMD has the hottest chips on earth" Adding that to the fact that I know there is a relationship between heat and reliability, it's not hard to understand why people (particularly corporations) will pay a premium for a reliability microprocessor. This bit o' logic forever vexes AMD fans, who scratch their heads in puzzlement on why everyone doesn't run out and buy AMD when it has comparable performance for less money at time zero. It's the same reason one might not buy a Yugo, even though it can probably get from the dealership to your house in the same amount of time as other more expensive cars. Talk to me in a few years. . .
To respect the Sandisk topic: flash companies also test their chips at high temperatures to simulate long-term data retention (and/or harsh conditions). OK, that's the best I could do!
Steve |
