Re: Re: Calling performance geeks

Hi Marcus, I'm very well aware of this. But to me, optimizing code to win CPU-cycles is one of the great pleasures I have in programming :). So I am and was fully aware that what I submitted here were probably those less interesting 90% of code that isn't used all the time. But I still enjoy finding these things, and I do think that a large amount of performance improvements in that less interesting 90% of the code, in the end can add up to something interesting (although I am aware that there probably isn't that much to improve on a code level, since I noticed the vast majority of the PHP codebase is already written in a pretty optimal way). So I knew my post here could be in vain, and maybe it still is. Feel free to apply the things I found, or don't :) No hard feelings from me either way. I enjoyed finding them :) Regards, Ron "Marcus Boerger" <helly@php.net> wrote in message news:1035317247.20060315021302@marcus-boerger.de... > Hello Ron, > > the first rule about optimizing is the 80-20 or 90-10 rule or whatever > you like more. Either way the idea is that the program spends more than > 90% of it's run-time in less than 10% of it's functions. This is a very > general idea but nontheless has proven right since people program. On > the contray this rule means that you should spend 90% of your optimizing > power in those 10%. Much of it identifying these 10%. Now guess you were > finding all the small things you aim for, you as must as increase he 90% > that have little to non influence on the overall run-time efficiency. > > > Just in case anybody wanted to hear this :-) > > > best regards > marcus > > Tuesday, March 14, 2006, 12:48:33 PM, you wrote: > > > It's probably not measurable, but a lot of small improvements like these may > > be measurable. I don't know how often this situation occurs per request, but > > if it happens for all apache config flags and php.ini flags, it may be worth > > improving. If an improvement is not measurable per request (but is > > measurable as an isolated case), is that a reason not to do it? I don't > > think so. But I'm not the one to judge :) So I'll just leave this up to the > > big kahuna's, and I don't really feel like something this small justifies > > any big discussions. > > > Ron > > > "Edin Kadribasic" <edink@emini.dk> schreef in bericht > > news:4416A492.9070909@emini.dk... > >> Ron Korving wrote: > >>> Oh right, okay. For a moment I thought you were talking about a > >>> variable-length case insensitive string compare based on this principle > >>> :) Your macro seems like a good idea to me. The function I was talking > >>> about was an apache-flag checking function, not php.ini, but I guess this > >>> could just as well be applied for php.ini. > >>> > >>> Like I said, the speedup is about 5x, so I think it's worth it. > >> > >> The speedup of a single function might as well be 5x, but how many time is > >> it called and can the performance increase be measured in benchmark test > >> such as the that was posted by Rasmus? > >> > >> Edin > > > > > Best regards, > Marcus