This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Ping^2: Unreviewed patch: GCSE redundant load after store elimination
- From: "Mostafa Hagog" <MUSTAFA at il dot ibm dot com>
- To: Gerald Pfeifer <gerald at pfeifer dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Wed, 15 Oct 2003 09:20:42 +0200
- Subject: Re: Ping^2: Unreviewed patch: GCSE redundant load after store elimination
>I noticed that several lines are too long (and wrap).
I agree. The problem is that there are long lines in the sources
so any patch can have these long lines. that's why I am
sending it as an attachment.
>Please do not attach patches as MIME type application/OCTET-STREAM which
>denotes _binary_ data which most mailers do not display right away.
I will try to configure my mail tool not to do that.
>> Also, we have seen that this patch bought us about 1% in SPECint on
>> powerpc-apple-darwin6.4.
>Nice. What's the compile-time impact?
>
>Gerald
I didn't measure that, but as you can see the only additional overhead
is considering stores, inserting them to the hash-table, and some
additional
redundancies. So I guess this is not a mentionable compile time impact.
However, to get to that improvement one should also increase the maximum
GCSE passes to 3, this maybe of an impact to the compile time.
Mostafa.