Path: utzoo!utgpu!watmath!uunet!tut.cis.ohio-state.edu!mailrus!rutgers!labrea!decwrl!purdue!bu-cs!mirror!necntc!ima!haddock!eli
From: eli@haddock.ima.isc.com (Elias Israel)
Newsgroups: gnu.gcc
Subject: Re: How good is '386 GCC?
Message-ID: <11530@haddock.ima.isc.com>
Date: 25 Jan 89 16:32:15 GMT
References: <3212@ima.ima.isc.com>
Reply-To: eli@haddock.ima.isc.com (Elias Israel)
Distribution: na
Organization: Interactive Systems, Boston
Lines: 17

I have gotten and compiled 1.32 on Interactive's 386/ix on my Compaq
386/25. It seems to work relatively well, and shows a 6% to 10% speed
increase over pcc in the drystone benchmark. I think it ought to be
possible to speed it up further and I am thinking about how to do
that. Any '386 hackers with ideas for common (or even uncommon)
peephole optimizations out there? I'd greatly appreciate the help.

Also, someone here has told me that some programs still cause the '386
GCC to abort, presumably because it has failed in an attempt to generate
a particular rtx or set of rtx's. I'd like to find these bugs and do
something about them, too.

Elias Israel		   | "Justice, n. A commodity which in more or
Interactive Systems Corp.  | less adulterated condition the State sells
Boston, MA		   | to the citizen as a reward for his allegiance,
..!ima!haddock!eli	   | taxes, and personal service."
			   |     -- Ambrose Bierce, _The Devil's Dictionary_