Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 23 Jun 2001 16:59:30 +0200
From:      "Karsten W. Rohrbach" <karsten@rohrbach.de>
To:        Terry Lambert <tlambert2@mindspring.com>
Cc:        Dag-Erling Smorgrav <des@ofug.org>, current@FreeBSD.ORG, obrien@FreeBSD.ORG
Subject:   Re: CPUTYPE warning
Message-ID:  <20010623165930.A14624@mail.webmonster.de>
In-Reply-To: <3B340F45.863D00A0@mindspring.com>; from tlambert2@mindspring.com on Fri, Jun 22, 2001 at 08:38:45PM -0700
References:  <xzppubyye0j.fsf@flood.ping.uio.no> <20010622192232.U64624@mail.webmonster.de> <3B340F45.863D00A0@mindspring.com>

next in thread | previous in thread | raw e-mail | index | archive | help

--ReaqsoxgOBHFXBhH
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

Terry Lambert(tlambert2@mindspring.com)@2001.06.22 20:38:45 +0000:
> "Karsten W. Rohrbach" wrote:
> >=20
> > btw, regarding gcc's -O2 optimization breakage on -2.95.x and improved
> > instrumentation of the new compiler kit, is there someone working on
> > getting gcc-3.0 into -current?
> >=20
> > ...yes *sigh* i know, 3.0 is _not_ stable, neither is -current ;-)
>=20
> Does it do tail-call optimization?  Last I checked, this
> was only supported under i986.

refer to http://www.gnu.org/software/gcc/gcc-3.0/features.html

---
General Optimizer Improvements:=20
    - Basic block reordering pass.=20
    - New if-conversion pass with support for conditional (predicated)
      execution.=20
    - New tail call and sibling call elimination optimizations.=20
    - New register renaming pass.=20
    - New (experimental) static single assignment (SSA) representation
      support.=20
    - New dead-code elimination pass implemented using the SSA representati=
on.=20
    - Global null pointer test elimination.=20
    - Global code hoisting/unification.=20
    - More builtins and optimizations for stdio.h, string.h and old BSD
      functions, as well as for ISO C99 functions.=20
    - New builtin __builtin_expect for giving hints to the branch predictor=
.=20
---

>=20
> It would be a _SERIOUS_ win, for a lot of stuff.

yes, i compiled it in a sandbox here just to look how some userspace
stuff for embedded use compiles and runs (let's call it "empiric
approach"), just to see whether it breaks some stuff or not and from
what i saw the files were very compact (due to dwarf2 debug symbols that
keep debug information very small), but the code also. with some tiny
binaries it came to an average of approx. 90% code size compared to
2.95.2 (stripped) binaries. also execution _seems_ to be quicker with
stuff that dynamically allocates mem and calls to shared objects/libs.

i did not yet have the time to look into the intrinsics but i was kinda
stunned about the improvements. next week when i am back in the office
i'll try to compile xfree4 just to see if it breaks something ;-)

have a nice weekend
/k

>=20
> -- Terry
>=20

--=20
> cd /pub; more beer
KR433/KR11-RIPE -- WebMonster Community Founder -- nGENn GmbH Senior Techie
http://www.webmonster.de/ -- ftp://ftp.webmonster.de/ -- http://www.ngenn.n=
et/
karsten&rohrbach.de -- alpha&ngenn.net -- alpha&scene.org -- catch@spam.de
GnuPG 0x2964BF46 2001-03-15 42F9 9FFF 50D4 2F38 DBEE  DF22 3340 4F4E 2964 B=
F46
Please do not remove my address from To: and Cc: fields in mailing lists. 1=
0x

--ReaqsoxgOBHFXBhH
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (FreeBSD)
Comment: For info see http://www.gnupg.org

iD8DBQE7NK7SM0BPTilkv0YRAgpBAJ45YxDe3vWNuyWEytAZ/vyrU2E5lwCfYfuL
U3xMMKazj31g+aQHDZ4G98g=
=+NvO
-----END PGP SIGNATURE-----

--ReaqsoxgOBHFXBhH--

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




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010623165930.A14624>