Re: PGCC package?
> Date: Sun, 12 Apr 1998 21:20:26 -0400 (EDT)
> From: Ossama Othman <othman@astrosun.tn.cornell.edu>
>
> The EGCS project has been attempting to recombine most or all of the gcc
> forks, including pgcc. Therefore, EGCS gcc should have most, if not all,
> of the optimizations available in the PGCC compiler. Some one please
> correct me if I am wrong (be nice about it :-)).
> Here is what the EGCS gcc info pages say about the optimizations:
>
> --------------------------------------------------
> Intel 386 Options
> -----------------
>
> These `-m' options are defined for the i386 family of computers:
>
> `-mcpu=CPU TYPE'
> Assume the defaults for the machine type CPU TYPE when scheduling
> instructions. The choices for CPU TYPE are: `i386', `i486',
> `i586' (`pentium'), `pentium', `i686' (`pentiumpro') and
> `pentiumpro'. While picking a specific CPU TYPE will schedule
> things appropriately for that particular chip, the compiler will
> not generate any code that does not run on the i386 without the
> `-march=CPU TYPE' option being used.
>
> `-march=CPU TYPE'
> Generate instructions for the machine type CPU TYPE. The choices
> for CPU TYPE are: `i386', `i486', `pentium', and `pentiumpro'.
> Specifying `-march=CPU TYPE' implies `-mcpu=CPU TYPE'.
I think my head is about to explode. I downloaded pgcc 1.0.2 a few
days ago and tested it against gcc 2.7.2.1 on Bytemark. I was running
it on a 256k-cache Pentium Pro and trying all the combinations of
optimisation switches that seemed to make sense the best I got out of
pgcc was slightly worse than that from 2.7.2.1.
Now, the FAQ for that pgcc said that it was based on egcs. It listed
various optimisation switches but didn't mention '-march'. After
reading your message I tried '-march=pentiumpro -O6' and it produced
the same code as '-mcpu=pentiumpro -O6'.
So.. is it worth me downloading egcs as well, in the hopes that
'-march=pentiumpro' will actually do something?
I was going to make a general remark about Pentium Pros here but
realised a) that it wasn't Debian-related, and b) that it was probably
flame bait. :-)
--
To UNSUBSCRIBE, email to debian-user-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: