Jim Prewett <[EMAIL PROTECTED]> writes:

> Hi,
> 
> I was thinking of changing some of my code (which is a very small class 
> that could easily be replaced with a small structure) in just this way.  
> 
> Is 19a going to make this not worth my time/effort?  I do like everything 
> in my system being CLOS classes, but I wasn't sure it is worth the 
> performance hit.  I'm guessing my case is pretty darn common, so now i'm 
> not sure if I should bother changing my code.  Thoughts?

In any case you should follow Ray's suggestion for profiling!

Nicolas.

P.S.: Yesterday I profiled a C++ program of a colleague of mine doing
similar things than yours.  It turned out (if I understood gprof's messages
correctly) that most of the time was spent in random number generation, and
that there was little hope in improving the performance if he did not want
to change the algorithm.

I also compared with CMUCL's random number generation, and it was about the
same speed as the Numerical Recipes code he used.  Thus, this kind of
program might even run with about the same speed when transformed into
Lisp.

Reply via email to