Re: genetic compilation

gneuner@dyn.EXTRACT.THIS.com (George Neuner)
20 Apr 2000 01:28:42 -0400

          From comp.compilers
Related articles
[4 earlier articles]
Re: genetic compilation Andy.Nisbet@cs.tcd.ie (Dr. Andy Nisbet) (2000-04-01)
Re: genetic compilation jean-luc.nagel@imt.unine.ch (Jean-Luc Nagel) (2000-04-01)
Re: genetic compilation cwfraser@microsoft.com (Chris Fraser) (2000-04-01)
Re: genetic compilation dvdeug@x8b4e53cd.dhcp.okstate.edu (2000-04-03)
Re: genetic compilation plakal@cs.wisc.edu (2000-04-03)
Re: Re: genetic compilation joachim.durchholz@halstenbach.com.or.de (Joachim Durchholz) (2000-04-14)
Re: genetic compilation gneuner@dyn.EXTRACT.THIS.com (2000-04-20)
| List of all articles for this month |
From: gneuner@dyn.EXTRACT.THIS.com (George Neuner)
Newsgroups: comp.compilers
Date: 20 Apr 2000 01:28:42 -0400
Organization: Dynamic ReSolutions, Inc.
References: 00-04-019 00-04-109
Keywords: architecture, optimize

On 14 Apr 2000 23:48:31 -0400, "Joachim Durchholz"
<joachim.durchholz@halstenbach.com.or.de> wrote:

>The story is even worse. Since a few years, microprocessors usually keep
>a record of what branches were taken or not taken in the past, and
>speculatively execute one or the other branch depending on what's
>recorded.

Not to mention the processors that speculatively execute *both*
branches and kill the results of the path not taken. The PowerPC does
this up to 3 levels for unresolved branches - ie. it can follow 8
different execution paths [to some depth] simultaneously.

George Neuner
Dynamic Resolutions, Inc.

Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.