Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 19 Jun 2002 00:50:18 +0300
From:      Giorgos Keramidas <keramida@ceid.upatras.gr>
To:        Mike Silbersack <silby@silby.com>
Cc:        Lucky Green <shamrock@cypherpunks.to>, freebsd-hackers@FreeBSD.ORG
Subject:   Re: LINT CPU features table
Message-ID:  <20020618215018.GD10528@hades.hell.gr>
In-Reply-To: <20020616223526.B13544-100000@patrocles.silby.com>
References:  <002d01c21573$d692ea50$0100a8c0@LUCKYVAIO> <20020616223526.B13544-100000@patrocles.silby.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On 2002-06-16 22:38 -0500, Mike Silbersack wrote:
> On Sun, 16 Jun 2002, Lucky Green wrote:
> > If only a few CPU's would benefit from the CPU-specific options,
> > creating a table of CPU options for those few CPU's should be all the
> > simpler. What I am I missing?
>
> IMHO, the performance benefits are so small, that it's best to not even
> concern people with making cpu-specific kernels.
>
> CPU-specific compiler options might actually make a difference, but those
> tend to create kernels that crash.

Or do not work flawlessly across hardware upgrades.  I have an
installation here at home that has gone through many hardware upgrades
(motherboard, cpu, or other vital parts) and has worked like a charm,
compiling worlds since 3.2-RELEASE from source.  Having a userland or
kernel that is 486-specific would have been a major PITA when I
changed the cpu to a Pentium, with a new PITA waiting at the next
corner, when I switched to a Celeron, etc.

I feel that it's very nice that "uname -p" and "uname -m" still print
"i386" in their output :-]

- Giorgos


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020618215018.GD10528>