public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc(refs/users/kubaneko/heads/histogram)] Clear profile in vectorizer.
@ 2023-05-01 22:39 Jan Hubicka
  0 siblings, 0 replies; only message in thread
From: Jan Hubicka @ 2023-05-01 22:39 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:8667da401c2968aab100c355c89e044b95314aca

commit 8667da401c2968aab100c355c89e044b95314aca
Author: Honza <jh@ryzen3.suse.cz>
Date:   Tue May 2 00:39:34 2023 +0200

    Clear profile in vectorizer.

Diff:
---
 gcc/tree-vect-loop-manip.cc | 6 ++++++
 gcc/tree-vect-loop.cc       | 3 +++
 2 files changed, 9 insertions(+)

diff --git a/gcc/tree-vect-loop-manip.cc b/gcc/tree-vect-loop-manip.cc
index 6aa3d2ed0bf..8eb2942b847 100644
--- a/gcc/tree-vect-loop-manip.cc
+++ b/gcc/tree-vect-loop-manip.cc
@@ -2941,6 +2941,10 @@ vect_do_peeling (loop_vec_info loop_vinfo, tree niters, tree nitersm1,
       /* Prolog iterates at most bound_prolog times, latch iterates at
 	 most bound_prolog - 1 times.  */
       record_niter_bound (prolog, bound_prolog - 1, false, true);
+      /* TODO: Update counters or at least release vectors.
+         We do not produce prologs on modern x86 targets we test on
+         so this is not important for proof of concept.  */
+      prolog->counters = NULL;
       delete_update_ssa ();
       adjust_vec_debug_stmts ();
       scev_reset ();
@@ -3077,6 +3081,8 @@ vect_do_peeling (loop_vec_info loop_vinfo, tree niters, tree nitersm1,
 	  /* -1 to convert loop iterations to latch iterations.  */
 	  record_niter_bound (epilog, bound - 1, false, true);
 	}
+      /* TODO: Compute epilog histograms or at least release the counters.  */
+      epilog->counters = NULL;
 
       delete_update_ssa ();
       adjust_vec_debug_stmts ();
diff --git a/gcc/tree-vect-loop.cc b/gcc/tree-vect-loop.cc
index ba068af264c..928079a7c9b 100644
--- a/gcc/tree-vect-loop.cc
+++ b/gcc/tree-vect-loop.cc
@@ -11139,6 +11139,9 @@ vect_transform_loop (loop_vec_info loop_vinfo, gimple *loop_vectorized_call)
 			  assumed_vf) - 1
 	 : wi::udiv_floor (loop->nb_iterations_estimate + bias_for_assumed,
 			   assumed_vf) - 1);
+  /* TODO: Fix profile after vectorization.  */
+  if (loop->counters)
+    loop->counters = NULL;
 
   if (dump_enabled_p ())
     {

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2023-05-01 22:39 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-05-01 22:39 [gcc(refs/users/kubaneko/heads/histogram)] Clear profile in vectorizer Jan Hubicka

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).