Subject: Re: improving ssh performance on sun4m systems
To: None <port-sparc@netbsd.org>
From: Charles Shannon Hendrix <shannon@widomaker.com>
List: port-sparc
Date: 03/16/2002 11:12:00
On Sun, Mar 17, 2002 at 12:15:15AM +1100, matthew green wrote:
>    
>    I was afraid of this after noting a few interesting test results.
>    
>    -mv8 enables the instruction set, -mtune=supersparc enables the
>    instruction ordering (anything else?), and -msupersparc does both.
>    
>    I think cpuflags should really issue the latter, because what it
>    does say, -mcpu=supersparc, is only the v8 instruction set.
> 
> 
> i think this is really a confusion caused by GCC itself;  it has
> slowly changed what it wants you to use over the years.  "-mv8"
> would probably not be accepted today, and is only retained for
> backwards compatibilty.  indeed, cpuflags should use
> 
> 	-mcpu=xxxsparc -mtune=xxxpsarc
> 
> for appropriate "xxx".


Right, the problem is that they really have two different types of
symbols: CPU architecture, and CPU architecture implementation.
v7, v8, sparclite, sparclet, and v9 are all architectures.  cypress,
supersparc, f930, and so on are implementations.

Because this isn't in the man pages, I think a lot of people assume that
things like mv8 give you everything, etc.

Tuning for an architecture generally only gives you instructions; you
have to specify an implementation to get things like instruction ordering.

I think they are moving to -mcpu because you can more clearly pick
what you want there.

-- 
UNIX/Perl/C/Pizza__________________________________shannon@widomaker.com