[patch] [wwwdocs] vectorizer webpage update

[Dorit Naishlos]

Thanks. Applied your comments, and committed.

dorit

Gerald Pfeifer <gerald@pfeifer.com> wrote on 15/03/2005 14:46:27:

> On Tue, 8 Mar 2005, Dorit Naishlos wrote:
> > I made further updates to the webpage. These include:
> > - unified "high level plan" section with "ToDo" section (into "high
level
> > plan").
> > - updated status for items in the "high level plan" section
> > - up to date todo list is under "Contributing" section
> > - moved items from "latest news" section to "Previous news and status"
> > - unified "old news" section with "status" section (into "Previous news
and
> > status")
> > - removed testcases lists
> >
> > Attached is "wwwdiff" - the diff assuming that the patch I sent last
week
> > was applied, and "wwwpatch.march8" which is the complete patch
(includes
> > all changes together).
> >
> > ok to install?
>
> Yes, thanks a lot (and sorry for the delay caused by my absence).
>
> There are some minor comments, which I'd like to ask you for
> consideration below, but nothing which should be more than ten
> minutes or so.
>
> Gerald
>
> Index: vectorization.html
> ===================================================================
> !   <dt>2005-03-01</dt>
> !     <dd>Vectorization projects for gcc 4.1: See
>
> GCC 4.1
>
>
> !       <a href=
> !             "http://gcc.gnu.org/wiki/Autovectorization%20Enhancements">
> !              http://gcc.gnu.org/wiki/Autovectorization%
> 20Enhancements</a>. </dd>
> !     <dd>Vectorization capabilities in gcc 4.0: See
>
> GCC 4.0
>
> +     <p>Current contributors to this project include
> +     Devang Patel, Ira Rosen, Keith Besaw, Leehod Baruch and Dorit
> Naishlos. This web page
> +     is managed by <a href="mailto:dorit@il.ibm.com">Dorit Naishlos
> <dorit@il.ibm.com></a>.
>
> "maintained" instead of "managed"?
>
> +         <li>Support certain operations on data-types that are not
> directly supported by
> +         a target, but yet vectorization is possible. For example,
> support data
> +         movements and bitwise operations on 64bit data types for
> altivec).</li>
>
> 64-bit
>
> +         <li>Improve debug information (mostly line-number information)
for
> +         vectorizer created code (see
>
> "code created by the vectorizer"?
>
> !             expression <tt>'a[2*i+1] = b[i]</tt>':</p>
>
> Please use <code> instead of <tt> here and in the following.
>
> !             <p>These operations (1 and 2 in the example) determine
>
> "examples 1 and 2"
>
> !             the data permutation that will be performed, however,
> !             they do not need to be directly vectorized. It maybe
>
> "may be" or "might be" (or "might be necessary" or "we might need to",
> which is active and preferrable to passive).
>
> !             required to mark such code to indicate that no vector
> !             code needs to be generated for it; instead, we may want
>
> ". Instead" will make this easier to parse.
>
>                   <li>Detect saturation idioms, and map them to the
> !                 appropriate vector operation. [initial support in
place.
> !       See <a href=
> !
"http://gcc.gnu.org/ml/gcc-patches/2004-12/msg00931.html">
> !                 http://gcc.gnu.org/ml/gcc-
> patches/2004-12/msg00931.html</a></li>
>
> Is there a ] missing?
>
> !                 <li>Improve aliasing analysis. [various gcc projects
> !       deal with improvements to alias analysis].</li>
>
> GCC
>