1. I think the biggest gain would come from the compiler optimizations one could do; that shouldn't be forgotten.
      posted by Jens Knutson at 12:11:37 AM on December 24, 2004  
  2. Wrong, they _should_ be forgotten

    how sad to see this day and age people still so insanely obsesed with pathetic performance optimizations
      posted by k at 03:42:07 PM on December 25, 2004  
  3. I do not quite agree.

    Even in the age of computers with incredible amounts of memory, fast as light itself, there are cases in which performance counts:

    1. Long and frequently running computations need to perform at some speed as the result will be needed someday.

    2. There are still environments (e.g. mobile phones, PDAs) that come with limited resources. For this devices, tight code is a must.

    3. Server-applications will be used concurrently - the tighter the code the more users can be served.

    4. As an aside, also think of the cycles you burn on unoptimized code, polluting the environment (so conveniently separated from you computer user)... ;)

    In summary, while premature optimizations are the root of at least a decent amount of evil, optimizations should not be forgotten. And if this optimizations do not interfere with read- or maintainability (like in "the compiler does it for you"), at least some coders will be glad to have them. I say, more power to them.
      posted by Andre Kloss at 10:06:40 AM on December 28, 2004  
  4. How much nonsense; this is clearly an early optimization; and by this time I thought everyone was suposed to know what that means.

    Performance works like this: you ignore it until you finish your project; then _if_ it's too slow; you _profile_; and then optimice until performance is _acceptable_; blindly handling the optimization stick around will only get you in a frking mess.
      posted by k at 05:49:20 PM on December 30, 2004