Path: gmd.de!Germany.EU.net!mcsun!uunet!cs.utexas.edu!usc!bloom-beacon.mit.edu!senator-bedfellow.mit.edu!athena.mit.edu!jfc
From: j...@athena.mit.edu (John F Carr)
Newsgroups: gnu.misc.discuss
Subject: Re: GCC support for xx86 [was Re: GPL not bypassed]
Date: 10 Jul 1993 03:20:20 GMT
Organization: Massachusetts Institute of Technology
Lines: 28
Distribution: gnu
Message-ID: <21lchk$rsi@senator-bedfellow.MIT.EDU>
References: <ABRAHAM.93Jul7011741@loke.iesd.auc.dk> <SCOTTM.93Jul62057 <SCT.93Jul9141158@ascrib.dcs.ed.ac.uk>
NNTP-Posting-Host: achates.mit.edu

In article <SCT.93Jul9141...@ascrib.dcs.ed.ac.uk>
	s...@dcs.ed.ac.uk (Stephen Tweedie) writes:

>Is there any indication that Intel may be willing to release the
>details of the 486 and 586 pipelines to the GCC maintainers?  This
>knowledge is essential to realising the full potential of those
>processors, and for the 586... sorry, Pentium, you can't get within
>30% of top performance without doing proper instruction scheduling.
>
>As far as I have read, Intel were planning on keeping this information
>restricted, and releasing it under license to compiler developers.
>If they are now taking a real financial interest in GCC, might they be
>willing to relax a little on this policy for the sake of getting
>proper code generation into GCC?

There was a lot of flaming about Pentium after the annoucement, and
some people called for a boycott of Intel because they wouldn't tell
people how to optimize for the Pentium.  It turned out that Intel did
provide sufficient information to optimize for Pentium.  The secret
chapter which was incorrectly reported to contain all information
needed to optimize in fact contains descriptions of privileged mode
features that are not required for compilers or even operating systems
(the features might be _useful_ to OS coders).



--
    John Carr (j...@athena.mit.edu)