Subject: Re: dynamic configuration (was Re: PR#4094)
To: Michael Richardson <>
From: Greywolf <>
List: tech-kern
Date: 10/07/2000 21:37:41
On Fri, 29 Sep 2000, Michael Richardson wrote:

#   I agree with CGD: appealing to tool chain compatibility is a
# red-herring. GCC is already the most widely ported and used C-compiler, to
# the point that I'm not really sure why Sun, HP and IBM still ship their
# own. I know that 3/4 of win32 environments that I've seen at customer sites
# are using MINGW32 now :-)

Because the compilers for Sun, HP and IBM a) are faster and b) generate
faster and tighter code than gcc.  We don't have a license for SUNWspro at
work anymore, I don't think, but I seem to remember it building things in
half the time it took gcc to do the same thing, and the default optimizing
pass put out better code (I could be mis-remembering but I don't think so).

I certainly know that when I had to migrate to gcc, I was quite frustrated
at how much slower it seemed to run.

GCC is a very generic solution.  It's widely ported, yes, but in that
wide portability seems to lie some performance hits at run-time.

[Much like the difference between FreeBSD and NetBSD on the i386, to
 my eyes.  FreeBSD is so machine-dependent on the i386, but from all
 accounts, including my own, it runs faster on the i386 than NetBSD
 does.  I prefer the heterogeneous ability of NetBSD, though, and the
 commonalities across platforms, so I run that instead.]

#   I must agree with Chris Torek's comments: merge bdevsv/cdevsw, but perhaps
# this is an unrelated task.

Only as long as we don't make Linux's mistake of eliminating the raw disk

BSD: Scalability Does Matter.