Re: using make to run binaries

[Andreas Maunz]

Hi all! Thanks to you all for the valuable and extensive answers.

> You should build the code with warnings turned on and read them and fix 
> them. Of course this should be the default condition but it's remarkable 
> how often it isn't, especially in fields like scientific programming 
> where SW development skills are a sideline. There are other things like 
> valgrind which can help but most likely the compiler can find the 
> problem if you let it.

Indeed we hadn't switched all warnings on. Now (with -Wall) it turned 
out that there were indeed two uninitialized floats in a vital part of 
the algorithm! In the light of this new finding the buggy results are 
now quite logical. It took me just 30 mins to track down the problem.
Code is running now smoothly with -O2. Lesson learned, I should say!

Greets, Andreas

--
http://www.maunz.de

               To know recursion, you must first know recursion.