This is the mail archive of the
mailing list for the GCC project.
Re: numerical instability and estimate-probability
- From: dewar at gnat dot com
- To: aoliva at redhat dot com, jh at suse dot cz
- Cc: gcc-pdo at atrey dot karlin dot mff dot cuni dot cz, gcc at gcc dot gnu dot org,mrs at windriver dot com, rth at redhat dot com
- Date: Sun, 18 Nov 2001 08:45:09 -0500 (EST)
- Subject: Re: numerical instability and estimate-probability
> On Thu, Nov 15, 2001 at 11:31:52AM -0800, mike stump wrote:
>> It is advantageous to have gcc emit the same code, on different hosts
>> for the same source code. This currently works.
> No, it doesn't. We have code that is affected by the undefined
> order of evaluation of function arguments. E.g.
> foo (gen_reg_rtx (...), gen_reg_rtx (...))
I see no good argument for damaging code quality by insisting on a canonical
order of evaluation here. The standard allows non-determinism here quite
deliberately for a good reason.