cycling through a number range

11.03.2009 - 21:36 (1 year, 5 months, 4 weeks, 2 days ago)

Filed under programming, .NET, geekdom, mathematics, TrivadisContent, C#

What I really wanted to do was to play a round of quakelive. But right now, quakelive is down (well, it's Beta, isn't it?). Quakelive pointed me to their twitter feed.

Next I wondered what other people are suffering from this outage and went to twitterfall, doing a search on "quake" and "quakelive". I found a completely unrelated entry on some Quake Engine code review of their networking code (sweet) and at the end a cute little piece of code which I wanted to try in C# straight away.

I hesitated, for my home PC has no Visual Studio. Alas, the trusty ol' csc is part of the .NET Framework runtime on Windows and after adding the relevant path to the %PATH% I was happily programming away with Notepad (by now you don't need syntax completion, do you? ;)

With the following code you can count up an index and let it roll over once it hits a limit and start with 0 again. Once you "get" the code you will understand that limit can be any number of the form 2n-1... 
int limit = 1; int count = 0;

while (true) {
  System.Threading.Thread.Sleep(300);
  Console.WriteLine(count);
  count = (count + 1) & limit;
}

It really isn't a big deal, but it's a nice reminder of binary arithmetics and if you are a performance geek (insert statement of premature optimization etc. here) you will like the fact that this counter is consistently faster than doing it with a modulo (%) operation. In fact, up to 4ms...repeating it 10'000'000 times. ;)

Comments

configurator ( 11.03.2009 - 23:58 UTC )
What about 
while (true) {
   for (int count = 0; i <= limit; i++;) {
      System.Threading.Thread.Sleep(300);
      Console.WriteLine(count);
   }
}
Both more readable and faster.
Delirious master of here ( 13.03.2009 - 09:10 UTC )
I am sure you know that wasn't my point. It is interesting, though, how very different the "for" performance is in a debug build and release build. While the bit ANDing is a lot quicker in a debug build, the simple "for"-count outpaces the ANDing in a release build.

Post a comment