It's pretty amazing how much computation you can buy for less than a cup of coffee. For less than 20 cents (in quantity, perhaps) you can buy a chip that out-p

It's pretty amazing how much computation you can buy for less than a cup of coffee. For less than 20 cents (in quantity, perhaps) you can buy a chip that out-performs the personal computers available in the early 80s. Of course you have to add peripherals to bring it to true parity, but you can probably have a working board for about five bucks that'll run rings around an Apple II or a vintage PC. The keyboard and monitor are the most expensive components. Likewise, memory. Recently I was thinking about doing some optimization and reorganization of some data for a hardware management project, when I realized that the data, for the entire life of the project, would fit into the CACHE of the processor it runs on. Projecting out five or six years, it would always fit. I stopped optimizing. Most of the time, the most valuable resource is the time of the person involved. Shaving milliseconds of response time rarely matters, shaving an hour of dev time does. (There are big exceptions to this when you are resource-constrained, as in video games, or hardware environments that need to use minimal memory or cycles for cost reasons). Premature optimization still remains a great evil.