One more improvement has been done in
the "Computing very large Fibonacci numbers" algorithm.
Its new version can be seen
at http://groups.google.com/groups?selm...0uni-berlin.de
Iterators in the transform() algorithm have been replaced by pointers.
--------------------------------------
vector<unsigned long> vin1, vin2, vout;
// Stuff
Previous version:
transform (vin1.begin(), vin1.end(), vin2.begin(), vout.begin(), functor());
Current version:
transform (&vin1[0], &vin1[0] + vin1.size(), &vin2[0], &vout[0], functor());
--------------------------------------
Measuring the comparative performance.
=====================================
-------------------------------------
Windows 2000
Intel (R) Celeron (R) CPU 1.70 GHz
GNU g++ 3.3.1 (cygming special), MinGW
------------ Compilation ------------------
$ g++ [Optimize option] -mno-cygwin foo.cpp
-------------------------------------------
Computing Fibonacci[50000].
Comparative performance.
|-------------------------------------------
| | transform | transform |
| Optimization | with | with |
| | iterators | pointers |
|-----------------|------------------------|
| No optimization | 4.846 | 3.044 |
| Optimization O1 | 2.573 | 2.133 |
| Optimization O2 | 2.223 | 2.323 |
| Optimization O2 | 2.233 | 2.223 |
--------------------------------------------
--
Alex Vinokur
http://mathforum.org/library/view/10978.html
http://sourceforge.net/users/alexvn
