Patchwork [rs6000] Change default tuning to POWER7

login
register
mail settings
Submitter Peter Bergner
Date Feb. 10, 2011, 2:26 p.m.
Message ID <1297348008.23730.23.camel@otta>
Download mbox | patch
Permalink /patch/82608/
State New
Headers show

Comments

Peter Bergner - Feb. 10, 2011, 2:26 p.m.
This patch changes the default processor tuning for powerpc64-linux builds
to POWER7.  This passed bootstrapping and regtesting.  Ok for mainline?

Peter

	* config/rs6000/linux64.h (PROCESSOR_DEFAULT): Change to
	PROCESSOR_POWER7.
	(PROCESSOR_DEFAULT64): Likewise.
David Edelsohn - Feb. 10, 2011, 4:08 p.m.
On Thu, Feb 10, 2011 at 9:26 AM, Peter Bergner <bergner@vnet.ibm.com> wrote:
> This patch changes the default processor tuning for powerpc64-linux builds
> to POWER7.  This passed bootstrapping and regtesting.  Ok for mainline?
>
> Peter
>
>        * config/rs6000/linux64.h (PROCESSOR_DEFAULT): Change to
>        PROCESSOR_POWER7.
>        (PROCESSOR_DEFAULT64): Likewise.

Okay.

Thanks, David
Peter Bergner - Feb. 10, 2011, 5:31 p.m.
On Thu, 2011-02-10 at 11:08 -0500, David Edelsohn wrote:
> On Thu, Feb 10, 2011 at 9:26 AM, Peter Bergner <bergner@vnet.ibm.com> wrote:
> > This patch changes the default processor tuning for powerpc64-linux builds
> > to POWER7.  This passed bootstrapping and regtesting.  Ok for mainline?
> >
> > Peter
> >
> >        * config/rs6000/linux64.h (PROCESSOR_DEFAULT): Change to
> >        PROCESSOR_POWER7.
> >        (PROCESSOR_DEFAULT64): Likewise.
> 
> Okay.

Thanks, committed as revision 170015.

Peter
Gerald Pfeifer - Feb. 11, 2011, 11:02 p.m.
On Thu, 10 Feb 2011, Peter Bergner wrote:
> This patch changes the default processor tuning for powerpc64-linux 
> builds to POWER7.  This passed bootstrapping and regtesting.  Ok for 
> mainline?

Any chance you can update the release notes at 
http://gcc.gnu.org/gcc-4.6/changes.html accordingly?

Gerald
David Edelsohn - Feb. 15, 2011, 4:19 p.m.
On Fri, Feb 11, 2011 at 6:02 PM, Gerald Pfeifer <gerald@pfeifer.com> wrote:
> On Thu, 10 Feb 2011, Peter Bergner wrote:
>> This patch changes the default processor tuning for powerpc64-linux
>> builds to POWER7.  This passed bootstrapping and regtesting.  Ok for
>> mainline?
>
> Any chance you can update the release notes at
> http://gcc.gnu.org/gcc-4.6/changes.html accordingly?

Gerald,

Is the appended website patch okay?

Thanks, David

Index: changes.html
===================================================================
RCS file: /cvs/gcc/wwwdocs/htdocs/gcc-4.6/changes.html,v
retrieving revision 1.114
diff -c -p -r1.114 changes.html
*** changes.html        10 Feb 2011 01:01:42 -0000      1.114
--- changes.html        15 Feb 2011 16:18:41 -0000
*************** supported on several Xilinx Spartan and
*** 845,850 ****
--- 845,854 ----
        instructions.  If the VSX instruction set is available, you can now use
        the new builtin functions <code>vec_vsx_ld</code> and
<code>vec_vsx_st</code>
        which always generates the VSX memory instructions.</li>
+     <li>The GCC compiler on AIX now defaults to a process layout with a
+       larger data space allowing larger programs to be compiled.</li>
+     <li>The default processor scheduling model and tuning for PowerPC64
+       Linux and for AIX 6.1 and above now is POWER7.</li>
    </ul>

  <h3>S/390, zSeries and System z9/z10, IBM zEnterprise z196</h3>
Gerald Pfeifer - Feb. 16, 2011, 2:21 a.m.
On Tue, 15 Feb 2011, David Edelsohn wrote:
> Is the appended website patch okay?

This looks good, with one change:  I believe Linux should be GNU/Linux,
or is this really only refering to the kernel and not a distribution?

> +     <li>The GCC compiler on AIX now defaults to a process layout with a
> +       larger data space allowing larger programs to be compiled.</li>
> +     <li>The default processor scheduling model and tuning for PowerPC64
> +       Linux and for AIX 6.1 and above now is POWER7.</li>

And perhaps just say "GCC on AIX"?

In any case, nice changes to the compiler.


Thanks,
Gerald
David Edelsohn - Feb. 16, 2011, 2:08 p.m.
On Tue, Feb 15, 2011 at 9:21 PM, Gerald Pfeifer <gerald@pfeifer.com> wrote:
> On Tue, 15 Feb 2011, David Edelsohn wrote:
>> Is the appended website patch okay?
>
> This looks good, with one change:  I believe Linux should be GNU/Linux,
> or is this really only refering to the kernel and not a distribution?
>
>> +     <li>The GCC compiler on AIX now defaults to a process layout with a
>> +       larger data space allowing larger programs to be compiled.</li>
>> +     <li>The default processor scheduling model and tuning for PowerPC64
>> +       Linux and for AIX 6.1 and above now is POWER7.</li>
>
> And perhaps just say "GCC on AIX"?

I wanted to distinguish that the process layout refers to the compiler
itself, not to the applications produced by the compiler.

- David
Gerald Pfeifer - Feb. 16, 2011, 2:24 p.m.
On Wed, 16 Feb 2011, David Edelsohn wrote:
>>> +     <li>The GCC compiler on AIX now defaults to a process layout with a
>>> +       larger data space allowing larger programs to be compiled.</li>
>> And perhaps just say "GCC on AIX"?
> I wanted to distinguish that the process layout refers to the compiler
> itself, not to the applications produced by the compiler.

And you did say that.  My bad, the text is actually quite clear.

Gerald

Patch

Index: config/rs6000/linux64.h
===================================================================
--- config/rs6000/linux64.h	(revision 169982)
+++ config/rs6000/linux64.h	(working copy)
@@ -73,9 +73,9 @@  extern int dot_symbols;
 #endif
 
 #undef  PROCESSOR_DEFAULT
-#define PROCESSOR_DEFAULT PROCESSOR_POWER6
+#define PROCESSOR_DEFAULT PROCESSOR_POWER7
 #undef  PROCESSOR_DEFAULT64
-#define PROCESSOR_DEFAULT64 PROCESSOR_POWER6
+#define PROCESSOR_DEFAULT64 PROCESSOR_POWER7
 
 /* We don't need to generate entries in .fixup, except when
    -mrelocatable or -mrelocatable-lib is given.  */