r19405: the talloc speed test suite was not giving an accurate picture of the

cost of talloc versus malloc. The size parameter in the test suite was
constantly increasing, leading to a worst case for malloc. It is far
more common to have talloc calls of 100 bytes or lower, so change the
benchmark to reflect this.

This makes talloc look much worse - on my laptop I now get:

talloc: 5615164 ops/sec
malloc: 14337130 ops/sec

I'm working on improving that.
(This used to be commit db273ef4dd6ab58da57bb4b59fb1fd5568585ed8)

1 files changed, 18 insertions, 13 deletions

diff --git a/source4/lib/talloc/testsuite.c b/source4/lib/talloc/testsuite.c
index 0658a931d0..f51252e57b 100644
--- a/source4/lib/talloc/testsuite.c
+++ b/source4/lib/talloc/testsuite.c
@@ -791,6 +791,8 @@ static bool test_speed(void)
 {
 	void *ctx = talloc_new(NULL);
 	unsigned count;
+	const int loop = 1000;
+	int i;
 	struct timeval tv;
 
 	printf("test: speed [\nTALLOC VS MALLOC SPEED\n]\n");
@@ -799,11 +801,13 @@ static bool test_speed(void)
 	count = 0;
 	do {
 		void *p1, *p2, *p3;
-		p1 = talloc_size(ctx, count);
-		p2 = talloc_strdup(p1, "foo bar");
-		p3 = talloc_size(p1, 300);
-		talloc_free(p1);
-		count += 3;
+		for (i=0;i<loop;i++) {
+			p1 = talloc_size(ctx, loop % 100);
+			p2 = talloc_strdup(p1, "foo bar");
+			p3 = talloc_size(p1, 300);
+			talloc_free(p1);
+		}
+		count += 3 * loop;
 	} while (timeval_elapsed(&tv) < 5.0);
 
 	fprintf(stderr, "talloc: %.0f ops/sec\n", count/timeval_elapsed(&tv));
@@ -814,15 +818,16 @@ static bool test_speed(void)
 	count = 0;
 	do {
 		void *p1, *p2, *p3;
-		p1 = malloc(count);
-		p2 = strdup("foo bar");
-		p3 = malloc(300);
-		free(p1);
-		free(p2);
-		free(p3);
-		count += 3;
+		for (i=0;i<loop;i++) {
+			p1 = malloc(loop % 100);
+			p2 = strdup("foo bar");
+			p3 = malloc(300);
+			free(p1);
+			free(p2);
+			free(p3);
+		}
+		count += 3 * loop;
 	} while (timeval_elapsed(&tv) < 5.0);
-
 	fprintf(stderr, "malloc: %.0f ops/sec\n", count/timeval_elapsed(&tv));
 
 	printf("success: speed\n");