This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Code emitted was bloated with no optimisation.
- From: Richard Sandiford <rdsandiford at googlemail dot com>
- To: Umesh Kalappa <umesh dot kalappa0 at gmail dot com>
- Cc: Andrew Haley <aph at redhat dot com>, "gcc\ at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>
- Date: Fri, 11 Apr 2014 19:27:21 +0100
- Subject: Re: Code emitted was bloated with no optimisation.
- Authentication-results: sourceware.org; auth=none
- References: <CAGfacvRROXs0OrSHxLiyOC_Rq2VOxDK5Phk+ScYh0BLqQG8jiQ at mail dot gmail dot com> <5346B79E dot 50504 at redhat dot com> <877g6wdt3b dot fsf at sandifor-thinkpad dot stglab dot manchester dot uk dot ibm dot com> <CAGfacvS=BvUq9GObS_7O0E2O2a88xuBQ0RSk3pwbpd1VLuAUqA at mail dot gmail dot com>
Umesh Kalappa <umesh.kalappa0@gmail.com> writes:
> Richard ,
> Pmode is defined HImode and private target is 16 bit where int ,short
> and Pmode is defined HImode and long as SImode.
>
> Please do let me know if it requires more information on the target.
In that case I suggest you compile with -da and compare the .expand
dumps for the optimised and unoptimised code. If they both use the same
bad sequence, see (a) which pass manages to optimise it and (b) figure
out why the expand-time code is so bad. If the difference is already
present at expand time then presumably something is going wrong earlier.
Thanks,
Richard