This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: GCC 3.3 compile speed regression - AN ANSWER
- From: Devang Patel <dpatel at apple dot com>
- To: gcc at gcc dot gnu dot org
- Cc: Devang Patel <dpatel at apple dot com>
- Date: Tue, 11 Feb 2003 12:46:03 -0800
- Subject: Re: GCC 3.3 compile speed regression - AN ANSWER
I checked with some source , microsofts compiler, the intel compiler and an old lattice compiler, gcc is somewhere around 4-6 times slower than these compilers, now this was a really bad test from benchmarking perspective, but still Tubo-C and Turbo-C++ did a lot of lines/sek.
And we do not want to forget other alternative, Mac OS X developers have -- CW.
Possibly it is fastest of all. We have received complained about GCC compiler time
speed in terms of couple of hours vs less than 10 minutes, 45 minutes vs less than
2 minutes (and that is compared to GCC 2.95... add appropriate numbers for GCC 3.x).
Precompiled headers is one solution and we have gained lot of ground
using PFE and PCH but still ...
As it was mentioned earlier in this thread, BSD and Linux developer also care
GCC's compilation speed. But most of them, I think, compare GCC 3.x against 2.95
hence we (gcc developers) are lucky that they have not seen fastest alternative.
We are working very hard to gain ground, but meanwhile we can not lose ground
from 3.2 to 3.3. Zem has identified two patches and this thread has not seen any constructive
discussions about them (in this mail, I'm also adding extra noise without much value).
So lets try to fix these already identified patches, since they are responsible for compile
speed regression.
-Devang