[PATCH 3/6] sched/numa: Avoid task migration for small numa improvement

[Srikar Dronamraju <srikar@linux.vnet.ibm.com>]

If numa improvement from the task migration is going to be very
minimal, then avoid task migration.

specjbb2005 / bops/JVM / higher bops are better
on 2 Socket/2 Node Intel
JVMS  Prev    Current  %Change
4     200892  210118   4.59252
1     325766  313171   -3.86627


on 2 Socket/4 Node Power8 (PowerNV)
JVMS  Prev     Current  %Change
8     89011.9  91027.5  2.26442
1     211338   216460   2.42361


on 2 Socket/2 Node Power9 (PowerNV)
JVMS  Prev    Current  %Change
4     190261  191918   0.870909
1     195305  207043   6.01009


on 4 Socket/4 Node Power7
JVMS  Prev     Current  %Change
8     57651.1  58462.1  1.40674
1     111351   108334   -2.70945


dbench / transactions / higher numbers are better
on 2 Socket/2 Node Intel
count  Min      Max      Avg      Variance  %Change
5      12254.7  12331.9  12297.8  28.1846
5      11851.8  11937.3  11890.9  33.5169   -3.30872


on 2 Socket/4 Node Power8 (PowerNV)
count  Min      Max      Avg      Variance  %Change
5      4997.83  5030.14  5015.54  12.947
5      4791     5016.08  4962.55  85.9625   -1.05652


on 2 Socket/2 Node Power9 (PowerNV)
count  Min      Max      Avg      Variance  %Change
5      9331.84  9375.11  9352.04  16.0703
5      9353.43  9380.49  9369.6   9.04361   0.187767


on 4 Socket/4 Node Power7
count  Min      Max      Avg      Variance  %Change
5      147.55   181.605  168.963  11.3513
5      149.518  215.412  179.083  21.5903   5.98948

Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
---
Changelog v1->v2:
 - Handle trivial changes due to variable name change. (Rik Van Riel)
 - Drop changes where subsequent better cpu find was rejected for
   small numa improvement (Rik Van Riel).

 kernel/sched/fair.c | 23 ++++++++++++++++++-----
 1 file changed, 18 insertions(+), 5 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 5cf921a..a717870 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -1568,6 +1568,13 @@ static bool load_too_imbalanced(long src_load, long dst_load,
 }
 
 /*
+ * Maximum numa importance can be 1998 (2*999);
+ * SMALLIMP @ 30 would be close to 1998/64.
+ * Used to deter task migration.
+ */
+#define SMALLIMP	30
+
+/*
  * This checks if the overall compute and NUMA accesses of the system would
  * be improved if the source tasks was migrated to the target dst_cpu taking
  * into account that it might be best if task running on the dst_cpu should
@@ -1600,7 +1607,7 @@ static void task_numa_compare(struct task_numa_env *env,
 		goto unlock;
 
 	if (!cur) {
-		if (maymove || imp > env->best_imp)
+		if (maymove && moveimp >= env->best_imp)
 			goto assign;
 		else
 			goto unlock;
@@ -1643,16 +1650,22 @@ static void task_numa_compare(struct task_numa_env *env,
 			       task_weight(cur, env->dst_nid, dist);
 	}
 
-	if (imp <= env->best_imp)
-		goto unlock;
-
 	if (maymove && moveimp > imp && moveimp > env->best_imp) {
-		imp = moveimp - 1;
+		imp = moveimp;
 		cur = NULL;
 		goto assign;
 	}
 
 	/*
+	 * If the numa importance is less than SMALLIMP,
+	 * task migration might only result in ping pong
+	 * of tasks and also hurt performance due to cache
+	 * misses.
+	 */
+	if (imp < SMALLIMP || imp <= env->best_imp + SMALLIMP / 2)
+		goto unlock;
+
+	/*
 	 * In the overloaded case, try and keep the load balanced.
 	 */
 	load = task_h_load(env->p) - task_h_load(cur);
-- 
1.8.3.1