This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Cleanups for the m68k backend
On Wed, Jul 09, 2003 at 07:51:42PM +0200, Bernardo Innocenti wrote:
> On Wednesday 09 July 2003 15:18, Gunther Nikl wrote:
> > Thats what I have observed too: the overhead of BFD-based binutils
> > is very large :-/ For an already slow platform compared to current
> > standards thats a pain.
>
> How is that? On modern platforms, the time spent by as + ld is negligible
> when compared with cc1. How come the same proportion doesn't hold for slower
> machines?
That's true for most input files, but there are certain cases where "ld"
times become large. One example is machine generated input files with
many strings, or many identifiers and debugging turned on. This isn't
so much a fault in newer "ld", but simply that "ld" is doing more work
handling the SHF_MERGE sections emitted by gcc. Newer versions of "ld"
may also use more memory.
--
Alan Modra
IBM OzLabs - Linux Technology Centre