re PR libgomp/49490 (suboptimal load balancing in loops)

	PR libgomp/49490
	* omp-low.c (expand_omp_for_static_nochunk): Only
	use n ceil/ nthreads size for the first
	n % nthreads threads in the team instead of
	all threads except for the last few ones which
	get less work or none at all.

	* iter.c (gomp_iter_static_next): For chunk size 0
	only use n ceil/ nthreads size for the first
	n % nthreads threads in the team instead of
	all threads except for the last few ones which
	get less work or none at all.
	* iter_ull.c (gomp_iter_ull_static_next): Likewise.
	* env.c (parse_schedule): If OMP_SCHEDULE doesn't have
	chunk argument, set run_sched_modifier to 0 for static
	resp. 1 for other kinds.  If chunk argument is 0
	and not static, set value to 1.

From-SVN: r175315

4 files changed, 40 insertions, 14 deletions

diff --git a/libgomp/ChangeLog b/libgomp/ChangeLog
index 7d3fe22..1d93273 100644
--- a/libgomp/ChangeLog
+++ b/libgomp/ChangeLog
@@ -1,3 +1,17 @@
+2011-06-22  Jakub Jelinek  <jakub@redhat.com>
+
+	PR libgomp/49490
+	* iter.c (gomp_iter_static_next): For chunk size 0
+	only use n ceil/ nthreads size for the first
+	n % nthreads threads in the team instead of
+	all threads except for the last few ones which
+	get less work or none at all.
+	* iter_ull.c (gomp_iter_ull_static_next): Likewise.
+	* env.c (parse_schedule): If OMP_SCHEDULE doesn't have
+	chunk argument, set run_sched_modifier to 0 for static
+	resp. 1 for other kinds.  If chunk argument is 0
+	and not static, set value to 1.
+
 2011-05-19  Jakub Jelinek  <jakub@redhat.com>
 
 	PR c++/49043
diff --git a/libgomp/env.c b/libgomp/env.c
index 92fa8c3..0ca9a1c 100644
--- a/libgomp/env.c
+++ b/libgomp/env.c
@@ -1,4 +1,4 @@
-/* Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010
+/* Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010, 2011
    Free Software Foundation, Inc.
    Contributed by Richard Henderson <rth@redhat.com>.
 
@@ -108,7 +108,11 @@ parse_schedule (void)
   while (isspace ((unsigned char) *env))
     ++env;
   if (*env == '\0')
-    return;
+    {
+      gomp_global_icv.run_sched_modifier
+	= gomp_global_icv.run_sched_var != GFS_STATIC;
+      return;
+    }
   if (*env++ != ',')
     goto unknown;
   while (isspace ((unsigned char) *env))
@@ -129,6 +133,8 @@ parse_schedule (void)
   if ((int)value != value)
     goto invalid;
 
+  if (value == 0 && gomp_global_icv.run_sched_var != GFS_STATIC)
+    value = 1;
   gomp_global_icv.run_sched_modifier = value;
   return;
 
diff --git a/libgomp/iter.c b/libgomp/iter.c
index 9ec4dbd..cd9484a 100644
--- a/libgomp/iter.c
+++ b/libgomp/iter.c
@@ -1,4 +1,4 @@
-/* Copyright (C) 2005, 2008, 2009 Free Software Foundation, Inc.
+/* Copyright (C) 2005, 2008, 2009, 2011 Free Software Foundation, Inc.
    Contributed by Richard Henderson <rth@redhat.com>.
 
    This file is part of the GNU OpenMP Library (libgomp).
@@ -59,7 +59,7 @@ gomp_iter_static_next (long *pstart, long *pend)
      trip through the outer loop.  */
   if (ws->chunk_size == 0)
     {
-      unsigned long n, q, i;
+      unsigned long n, q, i, t;
       unsigned long s0, e0;
       long s, e;
 
@@ -74,11 +74,14 @@ gomp_iter_static_next (long *pstart, long *pend)
       /* Compute the "zero-based" start and end points.  That is, as
          if the loop began at zero and incremented by one.  */
       q = n / nthreads;
-      q += (q * nthreads != n);
-      s0 = q * i;
+      t = n % nthreads;
+      if (i < t)
+	{
+	  t = 0;
+	  q++;
+	}
+      s0 = q * i + t;
       e0 = s0 + q;
-      if (e0 > n)
-        e0 = n;
 
       /* Notice when no iterations allocated for this thread.  */
       if (s0 >= e0)
diff --git a/libgomp/iter_ull.c b/libgomp/iter_ull.c
index 1754e63..a393920 100644
--- a/libgomp/iter_ull.c
+++ b/libgomp/iter_ull.c
@@ -1,4 +1,4 @@
-/* Copyright (C) 2005, 2008, 2009 Free Software Foundation, Inc.
+/* Copyright (C) 2005, 2008, 2009, 2011 Free Software Foundation, Inc.
    Contributed by Richard Henderson <rth@redhat.com>.
 
    This file is part of the GNU OpenMP Library (libgomp).
@@ -60,7 +60,7 @@ gomp_iter_ull_static_next (gomp_ull *pstart, gomp_ull *pend)
      trip through the outer loop.  */
   if (ws->chunk_size_ull == 0)
     {
-      gomp_ull n, q, i, s0, e0, s, e;
+      gomp_ull n, q, i, t, s0, e0, s, e;
 
       if (thr->ts.static_trip > 0)
 	return 1;
@@ -75,11 +75,14 @@ gomp_iter_ull_static_next (gomp_ull *pstart, gomp_ull *pend)
       /* Compute the "zero-based" start and end points.  That is, as
 	 if the loop began at zero and incremented by one.  */
       q = n / nthreads;
-      q += (q * nthreads != n);
-      s0 = q * i;
+      t = n % nthreads;
+      if (i < t)
+	{
+	  t = 0;
+	  q++;
+	}
+      s0 = q * i + t;
       e0 = s0 + q;
-      if (e0 > n)
-	e0 = n;
 
       /* Notice when no iterations allocated for this thread.  */
       if (s0 >= e0)