This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [autovect]: Handle indirect_refs in compute_data_dependence_for_loop, adda few more simple dependence tests, and trakc dependence stats
- From: Dorit Naishlos <DORIT at il dot ibm dot com>
- To: Sebastian Pop <sebastian dot pop at cri dot ensmp dot fr>
- Cc: Daniel Berlin <dberlin at dberlin dot org>, gcc-patches at gcc dot gnu dot org
- Date: Tue, 30 Nov 2004 19:17:39 +0200
- Subject: Re: [autovect]: Handle indirect_refs in compute_data_dependence_for_loop, adda few more simple dependence tests, and trakc dependence stats
> Dorit, do I have super powers on this branch? ;-)
Not that I know of... Lets continue to use the same policy we used for
lno-branch.
Daniel - please commit your patch. I'll try to switch the vectorizer to use
the improved utilities as soon as I can.
thanks!
dorit
Sebastian Pop
<sebastian.pop@cr To: Daniel Berlin <dberlin@dberlin.org>
i.ensmp.fr> cc: gcc-patches@gcc.gnu.org, Dorit Naishlos/Haifa/IBM@IBMIL
Subject: Re: [autovect]: Handle indirect_refs in
30/11/2004 18:56 compute_data_dependence_for_loop, add a few more simple dependence tests, and
trakc dependence stats
On Tue, Nov 30, 2004 at 11:06:47AM -0500, Daniel Berlin wrote:
> 2004-11-29 Daniel Berlin <dberlin@dberlin.org>
>
> * tree-data-ref.c (dependence_stats): New structure.
> (analyze_indirect_ref): New function.
> (analyze_ziv_subscript): Track ziv tests using
dependence_stats.
> (get_number_of_iters_for_loop): New function.
> Replace repeated idiom to get number of real/estimated
iterations
> in loop with this function.
> (analyze_siv_subscript_cst_affine): Track siv stats using
> dependence_stats.
> No dependence exists if first overlap is outside of the loop.
> (analyze_subscript_affine_affine): No dependence exists if
first
> overlap of either variable is outside of the loop.
> (analyze_siv_subscript): Track siv stats using
dependence_stats.
> (analyze_miv_subscript): Ditto for miv.
> (analyze_overlapping_iterations): Track total number of tests
> using dependence_stats.
> (compute_affine_dependence): Print out failing reference if
affine
> test is unusable.
> (find_data_references_in_loop): Rewrite to handle indirect_ref
as
> well using analyze_indirect_ref.
> (compute_data_dependences_for_loop): Initialize stats
structure,
> print out stats if request.
I like that patch.
Dorit, do I have super powers on this branch? ;-)
Sebastian