This is the mail archive of the
mailing list for the GCC project.
- From: Mark Mitchell <mark at codesourcery dot com>
- To: Robert Dewar <dewar at adacore dot com>
- Cc: David Edelsohn <dje at watson dot ibm dot com>, Richard Henderson <rth at redhat dot com>,Daniel Berlin <dberlin at dberlin dot org>, Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>, gcc at gcc dot gnu dot org
- Date: Tue, 04 Jan 2005 14:27:34 -0800
- Subject: Re: -funsafe-loop-optimizations
- Organization: CodeSourcery, LLC
- References: <20041231211409.GA22814@atrey.karlin.mff.cuni.cz> <Pine.LNX.firstname.lastname@example.org> <20041231232501.GA16663@redhat.com> <200501010543.j015h3D33264@makai.watson.ibm.com> <41DAE4DF.email@example.com> <41DB1726.firstname.lastname@example.org>
Robert Dewar wrote:
Mark Mitchell wrote:
David Edelsohn wrote:
While we discuss whether this should be the default or enabled at
any optimization level, can we agree that users should be able to assert
with a commandline option that they want less strict induction variable
semantics? I hope that we can move forward with an option to address
these performance regressions and allow users to request this
when they *do* want it, along the lines of the draft patch.
I would certainly agree. I think the first thing to do is to get a
patch that implements the semantics we want, in the aggressive mode;
then, we can decide whether we want it to be the default at some
optimization level or not.
I agree except that "less strict induction variable semantics" is
not an acceptable semantic description, and I think it is essential
to have a clear simple semantic description of what this means.
So to me, the first step is a spec that describes "the semantics we
A patch is the second step, not the first!