[C++] another testcase for long compile times

["Peter Bienstman" <Peter dot Bienstman at rug dot ac dot be>]

Hi all,

Over the last weeks, compile time perfomance has improved quite a lot, but
for some cases, it still takes 10 times as long to compile as compared to
2.95.2.

More specifically, these are the timings for the file you can find at
http://allserv.rug.ac.be/~pbienst/gcc/S_scheme.ii.gz:

gcc 2.95.2:
> time make S_scheme.o
g++ -O3 -march=pentiumpro -ftemplate-depth-30  -ffast-math -funroll-loops
-malign-double -fstrict-aliasing -I/home/pbienst/blitz-0.5beta3
-I/usr/local/include/python2.0 -c S_scheme.cpp

real    4m32.364s
user    3m37.610s
sys     0m13.690s

>gcc -v
Reading specs from
/home/pbienst/gcc_Exp/bin/../lib/gcc-lib/i686-pc-linux-gnu/2.97/specs
Configured with: ../gcc/configure --prefix=/home/pbienst/gcc_Exp/
--enable-languages=c++,f77 : (reconfigured)
gcc version 2.97 20010103 (experimental)
> time make S_scheme.o
g++ -O3 -march=pentiumpro -ftemplate-depth-30  -ffast-math -funroll-loops
-malign-double -fstrict-aliasing -I/home/pbienst/blitz-20001030
-I/usr/local/include/python2.0 -c S_scheme.cpp
real    46m51.945s
user    28m54.220s
sys     1m51.920s


I already found 4 minutes a bit steep, but 40 minutes compile times makes
gcc cvs absolutely unuseable for practical work.

PS: without optimisation, gcc cvs ICEs on this file.

Kind regards,

-------------------------------------
Peter Bienstman
Department of Information Technology
INTEC/IMEC - Ghent University
St.-Pietersnieuwstraat 41
B-9000 Gent - Belgium
E-mail: Peter.Bienstman@rug.ac.be
Tel: +32 9 264 3445
Fax: +32 9 264 3593
-------------------------------------