--randy
: Hello,

: Does the gcc flag -m486 do anything good on a Pentium-based computer when
: compiling?

: /o.

: --
: Oskar Widerberg - <os...@surfchem.kth.se>
: Computer & Network Administrator - Institute for Surface Chemistry
: Opinions are my own, not my employer's

--
 Christopher Neufeld - B.Sc., M.Sc., A.B.D.   neuf...@physics.utoronto.ca
 Home page:  http://caliban.physics.utoronto.ca/neufeld/Intro.html
 "Don't edit reality for the sake of simplicity"

Christopher> In article <DLLEwv....@beaver.cs.washington.edu>, Randy
Christopher>    Has somebody been overloading the "good" operator
Christopher> again?

Besides that, he isn't even completely correct. Besides the rather
exorbitant padding, -m486 also directs the compiler to use more RISCy
code. I would imagine you would loose speed by compiling with -m386 on
a Pentium. Of course, you can also turn off the excessive alignment by
itself...

--
                 Joel N. Squire <squ...@colorado.edu>
                      .oO finger for PGP key Oo.

Does the gcc flag -m486 do anything good on a Pentium-based computer when
compiling?

/o.

--
Oskar Widerberg - <os...@surfchem.kth.se>
Computer & Network Administrator - Institute for Surface Chemistry
Opinions are my own, not my employer's

There was a discussion recently on the linux-kernel mailing list in which
Linus suggested using:

  -fno-strength-reduce -m486 -malign-loops=2 -malign-jumps=2 -malign-function=2

The "-fno-strength-reduce" flag is there to avoid a bug in gcc, the rest to use
gcc 486 optimizations avoiding the 486's excessive alignment.

The -malign-* flags will work only on gcc-2.7.0 and later.

You might want to add this switches to

/usr/lib/gcc-lib/i486-linux/2.7.[012]/specs    (choose your version)

under *cc1: and *cc1plus:

Hope it helps.

P.S. Is this in the FAQs ? If it isn't I think it should be.

--
Luca Lizzeri                                               Casa: +39.+2.8053884
l...@niche.telnetwork.it                                  Ufficio: +39.+2.4815525
                                                            Fax: +39.+2.4982660

| There was a discussion recently on the linux-kernel mailing list in which
| Linus suggested using:
|
|   -fno-strength-reduce -m486 -malign-loops=2 -malign-jumps=2 -malign-function=2
|
| The "-fno-strength-reduce" flag is there to avoid a bug in gcc, the rest to use
| gcc 486 optimizations avoiding the 486's excessive alignment.

We had the same discussion in the SMP mailing list,
"-fno-strength-reduce" seems to be required for any optimization
greater than -O, such as -O{2,4,6} whatever. This seems to have been
around for quite a while, and I wonder if anyone would comment on
why it hasn't been fixed in gcc. I assume that this is because the
bug is subtle and hard to find, but I guess it could be that once
found it's hard to fix.

Do *not* build an SMP kernel without this flag!
--
        -bill davidsen (david...@tmr.com)
"As a software development model, Anarchy does not scale well."
                -Dave Welch