Does egcs do performance regression tests?

[Iain McClatchie]

I have a 5,000+ line program that can be compiled with either g++ or
gcc.  Basically, it's ANSI C that has prototypes and casts for
everything.

The program, compiled gcc -O2, runs faster than the same program
compiled g++ -O2.  That seems the wrong way around -- I would expect
g++ to be faster, since C++ is a (slightly) more restrictive language.
I note that the only #ifndef __cplusplus statements surround routines
that also have inline versions defined -- gcc needs these when it
compiles at low optimization levels.

I could try to narrow down which bits of code run more slowly with
g++ -O2 than with gcc -O2, to help the compiler writers.  Before I
do that I would like to know to whom I should submit a performance bug.
To do me any good, either my coding style or the compiler will have to
change within four months.  So I suspect I might want to submit the
problem to egcs rather than gcc.

I've looked over your email archives, and I find no other performance
bugs submitted.  Is there a seperate forum for such things, or are you
folks swamped with functionality issues right now?  Note that the
discrepancy I have found is between g++ and gcc -- the solution may be
right at hand.

-Iain McClatchie
iainmcc@ix.netcom.com