home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.arch
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!wupost!udel!rochester!cantaloupe.srv.cs.cmu.edu!netnews-2.srv.cs.cmu.edu!moss
- From: moss@cs.cmu.edu (Eliot Moss)
- Subject: Re: DEC Alpha architecture issues
- In-Reply-To: tremblay@flayout.Eng.Sun.COM's message of 19 Nov 92 18:52:35 GMT
- Message-ID: <MOSS.92Nov20101727@CRAFTY.cs.cmu.edu>
- Sender: news@cs.cmu.edu (Usenet News System)
- Nntp-Posting-Host: crafty.fox.cs.cmu.edu
- Reply-To: moss@cs.cmu.edu
- Organization: Dept of Comp and Info Sci, Univ of Mass (Amherst)
- References: <1992Nov18.112407.2518@doug.cae.wisc.edu>
- <1992Nov18.191730.1044@meiko.com> <lgnojjINN627@exodus.Eng.Sun.COM>
- Date: Fri, 20 Nov 1992 15:17:27 GMT
- Lines: 20
-
- Regarding conditional moves, I think that branch prediction is not the only
- issue. Another issue is code density. A comparison followed by a conditional
- move is one instruction shorter than a comparison followed by a branch
- followed by an unconditional move. The number of instructions moved through
- the pipe must be at least two (unless you are lucky to have a situation where
- a compare and branch can fold into one instruction) and the conditional move
- approach gives you that, but with no pipeline "bubble". In fact, conditional
- instructions are easy to handle in a regular pipeline -- you just suppress the
- store in the last step, and you certainly know the outcome of the comparison
- by then. In a superscalar, where things get reordered, it is admittedly not
- quite so simple.
- --
-
- J. Eliot B. Moss, Associate Professor Visiting Associate Professor
- Department of Computer Science School of Computer Science
- Lederle Graduate Research Center Carnegie Mellon University
- University of Massachusetts 5000 Forbes Avenue
- Amherst, MA 01003 Pittsburgh, PA 15213-3891
- (413) 545-4206, 545-1249 (fax) (412) 268-6767, 681-5739 (fax)
- Moss@cs.umass.edu Moss@cs.cmu.edu
-