Add more stats.

src/ph.c

@@ -445,10 +445,11 @@ PH_Stats(const struct ph * const restrict ph,
 	stats->maxlen = ph->maxlen;
 
 	stats->h2buckets_min = UINT64_MAX;
+	stats->h2strings_min = UINT64_MAX;
 	for (unsigned i = 0; i <= ph->h1->mask; i++)
 		if (vbit_test(ph->collision, i)) {
 			struct hash *h2;
-			uint64_t sz;
+			uint64_t sz, nstrings = 0;
 
 			h2 = ph->tbl[i].h2;
 			CHECK_OBJ_NOTNULL(h2, HASH_MAGIC);
@@ -460,8 +461,22 @@ PH_Stats(const struct ph * const restrict ph,
 			if (sz > stats->h2buckets_max)
 				stats->h2buckets_max = sz;
 			stats->h2buckets_avg +=
-				(sz - stats->h2buckets_avg)
-				/ stats->collisions;
+				((double)sz - stats->h2buckets_avg)
+				/ (double)stats->collisions;
+
+			for (unsigned j = 0; j < sz; j++) {
+				if (h2->tbl[j] == UINT_MAX)
+					continue;
+				nstrings++;
+			}
+			AN(nstrings);
+			if (nstrings < stats->h2strings_min)
+				stats->h2strings_min = nstrings;
+			if (nstrings > stats->h2strings_max)
+				stats->h2strings_max = nstrings;
+			stats->h2strings_avg +=
+				((double)nstrings - stats->h2strings_avg)
+				/ (double)stats->collisions;
 		}
 }
 

src/ph.h

@@ -53,6 +53,9 @@ struct ph_stats {
 	uint64_t	h2buckets_min;
 	uint64_t	h2buckets_max;
 	double		h2buckets_avg;
+	uint64_t	h2strings_min;
+	uint64_t	h2strings_max;
+	double		h2strings_avg;
 };
 
 /*

src/tests/bench/bench_ph.c

@@ -332,6 +332,9 @@ main(int argc, char *argv[])
 	printf("%lu h2 min buckets\n", stats.h2buckets_min);
 	printf("%lu h2 max buckets\n", stats.h2buckets_max);
 	printf("%.1f h2 mean buckets\n", stats.h2buckets_avg);
+	printf("%lu h2 min strings\n", stats.h2strings_min);
+	printf("%lu h2 max strings\n", stats.h2strings_max);
+	printf("%.1f h2 mean strings\n", stats.h2strings_avg);
 
 	if (do_iters == 0)
 		exit(EXIT_SUCCESS);