Re: Default ISA/tuning flags for GCC, --enable-kernel= level for glibc

[Dan Horák]

Jakub Jelinek píše v Po 26. 01. 2009 v 11:15 +0100:
>  Hi!
>  
>  dist-f11-gcc44 currently has gcc 4.4.0-0.X packages, which are almost ready
>  for moving into dist-f11, but before doing that, I'd like to discuss
>  defaults.  gcc on most arches has a switch to select which ISA it can assume
>  (i.e. what minimum CPU will gcc compiled code run on) and a separate option
>  to tune for the same or other CPU.  Currently in Fedora the tuning is
>  I believe:
>  		ISA		 tuning
>  i386		-march=i386	 -mtune=generic
>  x86_64		-march=x86-64	 -mtune=generic
>  ppc		lowest		 -mtune=power6
>  ppc64		lowest		 -mtune=power6
>  s390		-march=g5	 -mtune=z9-109
>  s390x		-march=z900	 -mtune=z9-109
>  ia64		lowest		 -mtune=itanium2
>  sparc		-mcpu=v7	 -mtune=v7
>  sparcv9		-mcpu=v7	 -mtune=v7
>  sparc64		-mcpu=ultrasparc -mtune=ultrasparc
>  
>  plus redhat-rpm-config tweaks something (-fasynchronous-unwind-tables
>  and -mminimal-toc tweaking should IMHO remain in redhat-rpm-config):
>  i686		-march=i686	 -mtune=generic
>  sparc		-mcpu=v7	 -mtune=ultrasparc
>  sparcv9		-mcpu=ultrasparc -mtune=ultrasparc
>  
>  The question is, do we want to change anything on this for F11?
>  E.g. I believe Fedora hasn't shipped an i386 or i486 kernel for years,
>  and especially the -march=i386 vs. -march=i486 difference is significant
>  (-march=i386 doesn't have sufficient atomic instructions).  The only
>  -lpthread library we are using is NPTL and that doesn't support i386
>  anyway, so I think we should either use -march=i486 or -march=i586
>  (the latter has 8 byte compare and swap instruction) as the default,
>  when -march= isn't specified.
>  Similarly, I wonder what CPUs we want to support in Fedora for PowerPC,
>  if e.g. -march=power4 default wouldn't be possible.  On s390, is somebody
>  running Fedora on some 31-bit hw still (G5/G6), or could we safely default to
>  -march=z900 for -m31 as well and say -mtune=z10?  On sparc I guess 
There is no plan for 31-bit Fedora on s390 AFAIK, and regarding 64-bit
please take a look at https://bugzilla.redhat.com/show_bug.cgi?id=463830

>  we could just -mcpu=ultrasparc -mtune=ultrasparc always, given that
> the
>  32-bit support is being dropped. 
The general question is what should be the relation between GCC defaults
and parameters set in the rpm build macros.


		Dan

Default ISA/tuning flags for GCC, --enable-kernel= level for glibc

[Rahul Sundaram]

Dan Horák wrote:

>  There is no plan for 31-bit Fedora on s390 AFAIK, and regarding
> 64-bit
>  please take a look at https://bugzilla.redhat.com/show_bug.cgi?id=463830 
Permission seems to be denied to non Red Hat logins.

Rahul

Default ISA/tuning flags for GCC, --enable-kernel= level for glibc

[Jakub Jelinek]

On Mon, Jan 26, 2009 at 02:37:28PM +0100, Dan Horák wrote:
>  There is no plan for 31-bit Fedora on s390 AFAIK, and regarding
> 64-bit
>  please take a look at https://bugzilla.redhat.com/show_bug.cgi?id=463830 
If we are to use -march=z9-109 -mtune=z10 instead of -march=z900 -mtune=z10
for s390 -m31 and -m64, then releng is important.  Aren't the current
mainframes used for Fedora (and RHEL) building z990?  If yes, then this
can't be switched until they are z9-109 or z10 based.

>  > we could just -mcpu=ultrasparc -mtune=ultrasparc always, given
> that the
>  > 32-bit support is being dropped.
>  
>  The general question is what should be the relation between GCC defaults
>  and parameters set in the rpm build macros. 
Certainly redhat-rpm-config macros shouldn't be weaker than what we default
in gcc (say if gcc defaults to -march=i486 -mtune=generic, then
redhat-rpm-config shouldn't override this to -march=i386 -mtune=generic).

	Jakub

Default ISA/tuning flags for GCC, --enable-kernel= level for glibc

[Dan Horák]

Jakub Jelinek píše v Po 26. 01. 2009 v 14:51 +0100:
>  On Mon, Jan 26, 2009 at 02:37:28PM +0100, Dan Horák wrote:
>  > There is no plan for 31-bit Fedora on s390 AFAIK, and regarding 64-bit
>  > please take a look at https://bugzilla.redhat.com/show_bug.cgi?id=463830
>  
>  If we are to use -march=z9-109 -mtune=z10 instead of -march=z900 -mtune=z10
>  for s390 -m31 and -m64, then releng is important.  Aren't the current
>  mainframes used for Fedora (and RHEL) building z990?  If yes, then this
>  can't be switched until they are z9-109 or z10 based.
>   
Good question, that needs to be checked.


		Dan

Default ISA/tuning flags for GCC, --enable-kernel= level for glibc

[Bill Nottingham]

Jakub Jelinek (jakub(a)redhat.com) said: 
>  > There is no plan for 31-bit Fedora on s390 AFAIK, and regarding
> 64-bit
>  > please take a look at https://bugzilla.redhat.com/show_bug.cgi?id=463830
>  
>  If we are to use -march=z9-109 -mtune=z10 instead of -march=z900 -mtune=z10
>  for s390 -m31 and -m64, then releng is important.  Aren't the current
>  mainframes used for Fedora (and RHEL) building z990?  If yes, then this
>  can't be switched until they are z9-109 or z10 based. 
They appear to be z9-109.

Bill