home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.dsp
- Path: sparky!uunet!telebit!phr
- From: phr@telebit.com (Paul Rubin)
- Subject: Re: Motorola's gcc diffs?
- In-Reply-To: jeff@dsp.sps.mot.com's message of 23 Dec 92 23:21:31
- Message-ID: <PHR.92Dec23224947@napa.telebit.com>
- Sender: news@telebit.com
- Nntp-Posting-Host: napa.telebit.com
- Organization: Telebit Corporation; Sunnyvale, CA, USA
- References: <1992Dec21.221938.14602@cheops.qld.tne.oz.au> <ANTSU.92Dec22230440@vega.utu.fi>
- <BzqIoE.35o@world.std.com> <JEFF.92Dec23232131@dsp.sps.mot.com>
- Date: 23 Dec 92 22:49:47
- Lines: 19
-
- In article <JEFF.92Dec23232131@dsp.sps.mot.com> jeff@dsp.sps.mot.com (Jeff Enderwick) writes:
- > >you will find the complete...
- >This version seems to work all right, but it generates illegal instruction
- >sequences. I was informed by Motorola that there is a post-optimizer
- >that is required, and that this code was not made freely available.
- >Although I was able to hack up a simple peephole optimizer to get rid
- >of the illegal instructions, the gcc code is very inefficient.
- >I threw the whole thing away. Perhaps Motorola worked very hard
- >on their optimizer and it does miraculous things. Perhaps not.
-
- This is all news to me. The post-optimizer isn't required, and it isn't
- based on FSF source.
-
- Jeff Enderwick
- jeff@dsp.sps.mot.com
-
- What does the post-optimizer do? How much does it improve
- the code? Thanks.
-
-