This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
gcc compile-time performance
- From: Dara Hazeghi <dhazeghi at pacbell dot net>
- To: gcc at gcc dot gnu dot org
- Date: Wed, 15 May 2002 17:04:26 -0700
- Subject: gcc compile-time performance
Hello all,
Seeing as how gcc 3.1 is almost here (comments from Klaus Berger
notwishstanding), I thought it would be interesting to see how the C compiler
has been faring. One of the issues I don't often see discussed is compile
time performance. In my experience, gcc runtime performance (at least on x86)
ranges from acceptable to quite good. However, compile-time seems to be
becoming more of an issue.
Here are timings of building the gcc 2.7.2.3 C compiler (make LANGUAGES=c)
with various compilers on a 733MHZ RedHat 7.2 i686-pc-linux-gnu box.
with CFLAGS=-O0
compiler buildtime
gcc 2.7.2.3 0m34.060s
gcc 2.8.1 0m36.720s
gcc 2.90.29 0m37.270s
gcc 2.91.66 0m38.380s
gcc 2.95.3 0m43.710s
gcc 3.0.4 0m51.370s
gcc 3.1 (pre) 0m59.630s
-
intel cc 5.01 0m42.620s
intel cc 6.0 0m38.280s
with CFLAGS=-O2
compiler buildtime
gcc 2.7.2.3 0m56.840s
gcc 2.8.1 1m5.190s
gcc 2.90.29 1m9.300s
gcc 2.91.66 1m18.840s
gcc 2.95.3 1m42.240s
gcc 3.0.4 2m6.730s
gcc 3.1 (pre) 2m54.690s
-
intel cc 5.01 2m22.910s
intel cc 6.0 1m38.380s
Obiously you are free to draw your own conclusions ;-) Now far be it for me
(who haven't contributed a line of code) to criticize, but at least with
respect to the optimized build, the slowdown from 3.0 to 3.1 is considerable
(a 38% increase in compile time). So some food for thought...
Dara
P.S. Another potentially interesting fact is that gcc 3.0.4 takes 37 minutes
to bootstrap on the same machine, but gcc 3.1 (pre 20020510) takes 61 minutes
(ada disabled).