[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Remove all do_inplace* logic from C++
From: |
Olaf Till |
Subject: |
Re: Remove all do_inplace* logic from C++ |
Date: |
Thu, 16 May 2013 07:32:32 +0200 |
User-agent: |
Mutt/1.5.20 (2009-06-14) |
On Wed, May 15, 2013 at 11:36:46AM -0400, Jordi GutiƩrrez Hermoso wrote:
> <snip>
> Any objection to completely removing all of the inplace logic from C++?
What is if the lhs operand (m-code) contains, e.g., indexing
operations? E.g.
a(id1, id2, id3) += b;
against
a(id1, id2, id3) = a(id1, id2, id3) + b;
or even
c{id4, id5, id6}(id1, id2, id3) += b;
against
c{id4, id5, id6}(id1, id2, id3) = c{id4, id5, id6}(id1, id2, id3) + b;
or even more complicated?
Have you checked whether Octaves do_inplace code you speak of is able
to spare the double evaluation of the index operation? If it is, this
would be surely more than an optimizing compiler could achieve,
wouldn't it?
I do not mention structure field lookups because I remember heaving
read they are cached.
Olaf
--
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net
signature.asc
Description: Digital signature