This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
RE: OT: sparc, sparcv9, & sparc64 [Was: What about sparcv9-linux as target for configure?]
- From: Christian Jönsson <christian at j-son dot org>
- To: "'gcc'" <gcc at gcc dot gnu dot org>
- Date: Thu, 12 Sep 2002 21:35:51 +0200
- Subject: RE: OT: sparc, sparcv9, & sparc64 [Was: What about sparcv9-linux as target for configure?]
Sorry, it wasn't my intention to whine, thank you for your
clarification.
Cheers,
/ChJ
-----Original Message-----
From: gcc-owner@gcc.gnu.org [mailto:gcc-owner@gcc.gnu.org] On Behalf Of
Robert Schiele
Sent: Thursday, September 12, 2002 9:29 PM
To: gcc
Subject: Re: OT: sparc, sparcv9, & sparc64 [Was: What about
sparcv9-linux as target for configure?]
On Thu, Sep 12, 2002 at 07:33:29PM +0200, Christian Jönsson wrote:
> uhm, I'm actually a bit confused here... as I understand it, and right
> now it seems I understand this wrong, is that 'sparc' is the same as
> sparcv7 actually with for example sun4c (any more?). now, there would
> naturally be a sparcv8 also perhaps (like sun4d and sun4m) and there
> is the sparcv9 like the sun4u. Now, where does sparc64 come in? are we
> saying sparc64 *is* sparcv9? nah... then sparc32 would be sun4{c,d,m},
> right?
>
> and, looking at how redhat/rawhide/aurora sets it's compier options:
>
> optflags: sparc -O2 -m32 -mtune=ultrasparc
> optflags: sparcv9 -O2 -m32 -mcpu=ultrasparc
> optflags: sparc64 -O2 -m64 -mcpu=ultrasparc
>
> I'd say sparc64 and sparcv9 is not the same...
>
> and btw aurora's suggestion to have
>
> optflags: sparcv8 -O2 -m32 -mcpu=v8 -mtune=ultrasparc
>
> is perhaps coming also (mainly for multiplication heavy stuff like
> openssl and glibc or rather libm there).
I know that this is not consistent in all places. I also know that rpm
uses these aliases vor compiler setup. But I also know---and you should
consider that---that within Solaris a 64 bit compiler is built, when you
configure with sparcv9. There was a discussion about changing this to
32 bit on this list once. It was even changed for a short period of
time. But finally a decision was made to put this back to 64 bit.
And a fact is that only gcc for Solaris accepts sparcv9, where this is
the same as sparc64.
So you have multiple options:
- You could whine about the current situation.
- You could improve it by providing a patch that implements a better
commonly accepted solution.
- You could be pragmatic and just configure with sparc64 if you want a
64 bit compiler and build with just sparc for a 32 bit compiler and
setup the optimization with -mcpu=ultrasparc.
Robert
--
Robert Schiele Tel.: +49-621-181-2517
Dipl.-Wirtsch.informatiker mailto:rschiele@uni-mannheim.de