home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!uknet!yorkohm!minster!mjl-b
- From: mjl-b@minster.york.ac.uk
- Newsgroups: comp.arch
- Subject: Re: Optimizing Old Code for i586
- Message-ID: <721415010.2903@minster.york.ac.uk>
- Date: 10 Nov 92 17:03:30 GMT
- References: <1992Nov6.194508.9171@tc.cornell.edu>
- Reply-To: mjl-b@minster.york.ac.uk (Mathew Lodge)
- Organization: Department of Computer Science, University of York, England
- Lines: 31
-
- In article <1992Nov6.194508.9171@tc.cornell.edu> elan@tasha.cheme.cornell.edu (Elan Feingold) writes:
- >OK, I don't read this group, so I don't know if this has been talked about
- >much already, but I woke up last night at about 4AM in a cold sweat
- >and with this wierd idea:
- >
- >Write a program that takes a .exe file and performs the following analysis
- >on it:
- >
- > a) Break up the code into windows, bounded by branches.
- > b) Rearrange the code within the windows following the i586
- > idea of being able to execute two instructions in a cycle if
- > there are no dependancies. The idea is that since the 586 only
- > "looks ahead" one instruction, one helps the process by spending
- > time reorganizing the code in the .exe to maximize the chance that
- > the i586 will be able to execute two consecutive instructions concurrently.
- >
- >Is this a really stupid idea?
- >Would the performance increase be anywhere noticiable?
- >Will compilers for the i586 probably perform this optimization?
-
- If it turns out to be a useful win, then probably, yes. However, code motion
- can cause problems for error handling, and particularly for exception
- handling (in those languages that support it -- eg Ada)
-
- >| Elan Feingold | .sig currently undergoing renovation... |
-
- Mat
-
- | Mathew Lodge | "I don't care how many times they go |
- | mjl-b@minster.york.ac.uk | up-tiddly-up-up. They're still gits." |
- | Langwith College, Uni of York, UK | -- Blackadder Goes Forth |
-