cc/gcc/egcs kernel compilation options?

cc/gcc/egcs kernel compilation options?

[Barrie Spence]

As a PPC newbie, why do we need/use "-msoft-float"? Is this 
to handle some processor implementations that don't have 
floating-point hardware? (and if so, does it have any overhead 
on the 603/604/G3 Macs?). If it does have some cost, could it 
not be enabled in the same way as on Intel?

Next question, is there any advantage to building a kernel
with "-mcpu=604 -mtune=604"? (or whatever is most appropriate).

Thanks,
	Barrie


--
Barrie Spence			Sanity Clause? There is no Sanity Clause
Home: barrie@calvin.demon.co.uk		Telephone +44 1506 442304
Play: b.spence@computer.org		

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: cc/gcc/egcs kernel compilation options?

[Anirudh Joshi]

	we use -msoft-float (notably for kernel builds) becuase the kernel
has problems with fpu ops and things go crazy, build without -msoft-float
and you'll find out :)
	i'm not sure if -mcpu=60X will do much good for kernel builds,
but for most others it should be fine, all it does is add a bunch of
'good' optimizations suggested.



ani




On Thu, 1 Jul 1999, Barrie Spence wrote:

> 
> As a PPC newbie, why do we need/use "-msoft-float"? Is this 
> to handle some processor implementations that don't have 
> floating-point hardware? (and if so, does it have any overhead 
> on the 603/604/G3 Macs?). If it does have some cost, could it 
> not be enabled in the same way as on Intel?
> 
> Next question, is there any advantage to building a kernel
> with "-mcpu=604 -mtune=604"? (or whatever is most appropriate).
> 
> Thanks,
> 	Barrie
> 
> 
> --
> Barrie Spence			Sanity Clause? There is no Sanity Clause
> Home: barrie@calvin.demon.co.uk		Telephone +44 1506 442304
> Play: b.spence@computer.org		

> 


[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: cc/gcc/egcs kernel compilation options?

[Daniel Jacobowitz]

On Thu, Jul 01, 1999 at 10:49:57PM +0000, Barrie Spence wrote:
> 
> As a PPC newbie, why do we need/use "-msoft-float"? Is this 
> to handle some processor implementations that don't have 
> floating-point hardware? (and if so, does it have any overhead 
> on the 603/604/G3 Macs?). If it does have some cost, could it 
> not be enabled in the same way as on Intel?
> 
> Next question, is there any advantage to building a kernel
> with "-mcpu=604 -mtune=604"? (or whatever is most appropriate).

Floating point operations in the kernel are pretty rare, and thus do
not need to be optimized for speed; soft-float is thus acceptable.  The
advantage is that we don't need to save/restore the FPU registers and
state nearly as often.  And I do not know why x86 does not have this
issue.

Dan

/--------------------------------\  /--------------------------------\
|       Daniel Jacobowitz        |__|        SCS Class of 2002       |
|   Debian GNU/Linux Developer    __    Carnegie Mellon University   |
|         dan@debian.org         |  |       dmj+@andrew.cmu.edu      |
\--------------------------------/  \--------------------------------/

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]

Re: cc/gcc/egcs kernel compilation options?

[Michael Meissner]

On Wed, Jul 07, 1999 at 01:03:17PM -0400, Daniel Jacobowitz wrote:
> 
> On Thu, Jul 01, 1999 at 10:49:57PM +0000, Barrie Spence wrote:
> > 
> > As a PPC newbie, why do we need/use "-msoft-float"? Is this 
> > to handle some processor implementations that don't have 
> > floating-point hardware? (and if so, does it have any overhead 
> > on the 603/604/G3 Macs?). If it does have some cost, could it 
> > not be enabled in the same way as on Intel?
> > 
> > Next question, is there any advantage to building a kernel
> > with "-mcpu=604 -mtune=604"? (or whatever is most appropriate).
> 
> Floating point operations in the kernel are pretty rare, and thus do
> not need to be optimized for speed; soft-float is thus acceptable.  The
> advantage is that we don't need to save/restore the FPU registers and
> state nearly as often.  And I do not know why x86 does not have this
> issue.

The -msoft-float is there to prevent the compiler from converting some double
word moves into load floating/store floating combination.  The x86 compiler
does not allow values with integer modes (QImode, HImode, SImode, etc.) in fp
registers.  The powerpc can't do that since it would break int<->fp
conversions.

-- 
Michael Meissner, Cygnus Solutions
PMB 198, 174 Littleton Road #3, Westford, Massachusetts 01886
email: meissner@cygnus.com	phone: 978-486-9304	fax: 978-692-4482

[[ This message was sent via the linuxppc-dev mailing list.  Replies are ]]
[[ not  forced  back  to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. Please check http://lists.linuxppc.org/ ]]
[[ and http://www.linuxppc.org/ for useful information before posting.   ]]