This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: build and install of egcs-1.0.3a
- To: rrh at dial dot pipex dot com (Robley Hall)
- Subject: Re: build and install of egcs-1.0.3a
- From: Joe Buck <jbuck at synopsys dot com>
- Date: Fri, 5 Jun 98 17:22:22 PDT
- Cc: egcs at cygnus dot com
> However egcs is slower than my old gcc-2.7, as well as generating
> significantly larger programs. (Maybe this is expected, but I'll pass
> along the details anyway. Discard if you already know this.)
The example you show is of two extremely small programs. I believe
the reason you are seeing a size increase is because there is more
stuff in libgcc.a, which is statically linked.
1.0.3:
> ******* output from time make **********
> 38.03user 0.90system 0:39.29elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (0major+0minor)pagefaults 0swaps
> ******* output from gcc -v ************
> Reading specs from /usr/local/lib/gcc-lib/i586-pc-linux-gnulibc1/egcs-2.90.29/specs
> gcc version egcs-2.90.29 980515 (egcs-1.0.3 release)
> ******* output from ls -l **************
> -rwxr-xr-x 1 rrh users 93226 Jun 4 15:25 bzip2*
> -rwxr-xr-x 1 rrh users 34397 Jun 4 15:25 bzip2recover*
>
> <<< - - - - - end of output from make using egcs - - - - - >>>
> ******* output from time make **********
> 30.80user 1.01system 0:31.84elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k
> 0inputs+0outputs (0major+0minor)pagefaults 0swaps
> ******* output from gcc -v ************
> Reading specs from /usr/lib/gcc-lib/i486-linux/2.7.2.1/specs
> gcc version 2.7.2.1
> ******* output from ls -l **************
> -rwxr-xr-x 1 rrh users 73361 Jun 4 15:11 bzip2*
> -rwxr-xr-x 1 rrh users 13780 Jun 4 15:11 bzip2recover*
Your data shows an approximate 20K size increase for both programs.
I believe that this is a fixed cost, roughly (meaning it's not going
to be significant for large programs) and that it comes from libgcc.a,
though I'm not certain.
As for 39 sec vs 32 sec, we'd need to see more data, like the most
significant datum of all: which bzip2 executable runs faster and by
how much? (I'm not expecting much for egcs at this stage, though
improvements being worked on should be significant).