This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: if/then/else hint ?

From: Paul Koning <pkoning at equallogic dot com>
To: jamie at shareable dot org
Cc: rys at epaibm dot rtpnc dot epa dot gov, gcc at gcc dot gnu dot org
Date: Thu, 8 Jan 2004 13:22:13 -0500
Subject: Re: if/then/else hint ?
References: <3FFD9840.9030000@trex.rtpnc.epa.gov><20040108180526.GB12095@mail.shareable.org>

>>>>> "Jamie" == Jamie Lokier <jamie@shareable.org> writes:

 Jamie> When a branch is not already in the branch target buffer (BTB;
 Jamie> a small cache of previously taken brances to predict them
 Jamie> subsequently), a Pentium guesses that a forward branch is not
 Jamie> taken and a backward branch is taken.  So you're right to
 Jamie> observe that the code is not optimal.

Interestingly enough, that is true also for many other processors, but
not all.  The Moto 68040 always predicts branches to be taken no
matter whether they are forward or backward branches.

       paul

Follow-Ups:
- Re: if/then/else hint ?
  - From: Richard Earnshaw

References:
- if/then/else hint ?
  - From: Greg Smith
- Re: if/then/else hint ?
  - From: Jamie Lokier

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]