🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: gnu.gcc.help
Article #298

Re: gcc2.2.2 and X11R5

#298
From: Casper H.S. Dik
Date: Sat, 29 Aug 1992 22:11
41 lines
2431 bytes
brendan@cygnus.com (Brendan Kehoe) writes:

>In article <1992Aug20.165136.2723@ennews.eas.asu.edu> scowen@wfpc3.la.asu.edu (Paul A. Scowen) writes:

>   My recommendation to all people out there attempting this is DON'T USE GCC
>   vers 2.2.2 TO BUILD X!!!  It does not appear to have what it takes.

>I built it with gcc2, and it works wonderfully.  X11R5 patched to 16
>(I think it was), and the patches in README.X11 applied from the gcc
>2.2.2 distribution.  No other changes.  I suggest you start from
>scratch, since you obviously configured something wrong.

>I strongly recommend you *do* use it; the performance improvements are
>noticeable, in addition to the size reduction (as compared to gcc1).

Gcc 1.x generated extremely poor code for SPARC machines. It didn't
even come close to Sun's bundled C compiler (30%-100% slower, in
my experience). The other reason not to use GCC 1.x for SPARC
based machines was the fact that gcc 1.x wasn't fully call
compatible.

In gcc 2.x most of this has changed. The code generated is now
about the same speed to somewhat faster than the code generated by
Sun's bundled compielrs. However, one point of concern remains.
Gcc generates lots of calls to functions from libgcc.a for
certain primitives. This means that you cannot have users
choose between compilers when compiling X programs, but force
them to use gcc or go to some extra length to include libgcc.a on the
final link command line.

The default gcc setup for X will compile the libraries with
Sun's cc, the special setup for Gcc2 will compile the libraries
gcc, as it now understands -fpic. I would recommend against that,
for the above mentioned reasons. Most of the performance
benefits can be had from speeding up the server, in my opinion.

Casper

-- 
						|	Casper H.S. Dik
						|	casper@fwi.uva.nl
Technical Details
Message-ID: <1992Aug29.221146.6227@fwi.uva.nl>
Path: arc1-ca.pugleaf.net!sparky!uunet!math.fu-berlin.de!news.netmbx.de!Germany.EU.net!mcsun!sun4nl!fwi.uva.nl!casper
References: <1992Aug20.125042.13301@midway.uchicago.edu> <1992Aug20.163706.1779@ennews.eas.asu.edu> <1992Aug20.165136.2723@ennews.eas.asu.edu> <34569@hoptoad.uucp>

Show Full Headers
Complete Article Headers
Newsgroups: gnu.gcc.help
Path: arc1-ca.pugleaf.net!sparky!uunet!math.fu-berlin.de!news.netmbx.de!Germany.EU.net!mcsun!sun4nl!fwi.uva.nl!casper
From: casper@fwi.uva.nl (Casper H.S. Dik)
Subject: Re: gcc2.2.2 and X11R5
Message-ID: <1992Aug29.221146.6227@fwi.uva.nl>
Sender: news@fwi.uva.nl
Nntp-Posting-Host: adam.fwi.uva.nl
Organization: FWI, University of Amsterdam
References: <1992Aug20.125042.13301@midway.uchicago.edu>  <1992Aug20.163706.1779@ennews.eas.asu.edu>  <1992Aug20.165136.2723@ennews.eas.asu.edu> <34569@hoptoad.uucp>
Date: Sat, 29 Aug 1992 22:11:46 GMT
Lines: 41