lib/ccan/htable, strset: benchmarking tools.

This lets us compare hash table vs. strset vs. the example
implementation of critbit trees.

cbspeed 100 runs, min-max(avg):
#01: Initial insert:   236-245(237)
#02: Initial lookup (match):   180-186(180)
#03: Initial lookup (miss):   171-185(172)
#04: Initial lookup (random):   441-457(444)
#05: Initial delete all:   127-132(128)
#06: Initial re-inserting:   219-225(220)
#07: Deleting first half:   101-104(102)
#08: Adding (a different) half:   158-162(159)
#09: Lookup after half-change (match):   202-207(203)
#10: Lookup after half-change (miss):   217-222(218)
#11: Churn 1:   297-302(299)
#12: Churn 2:   297-305(300)
#13: Churn 3:   301-308(303)
#14: Post-Churn lookup (match):   189-195(190)
#15: Post-Churn lookup (miss):   189-193(190)
#16: Post-Churn lookup (random):   499-513(503)

speed 100 runs, min-max(avg):
#01: Initial insert:   211-218(212)
#02: Initial lookup (match):   161-166(162)
#03: Initial lookup (miss):   157-162(158)
#04: Initial lookup (random):   452-460(454)
#05: Initial delete all:   126-135(127)
#06: Initial re-inserting:   193-201(194)
#07: Deleting first half:   99-107(99)
#08: Adding (a different) half:   143-190(144)
#09: Lookup after half-change (match):   183-195(184)
#10: Lookup after half-change (miss):   197-203(198)
#11: Churn 1:   271-278(274)
#12: Churn 2:   280-287(282)
#13: Churn 3:   277-285(279)
#14: Post-Churn lookup (match):   171-175(171)
#15: Post-Churn lookup (miss):   174-178(175)
#16: Post-Churn lookup (random):   525-552(528)

stringspeed 100 runs, min-max(avg):
#01: Initial insert:   300-343(308)
#02: Initial lookup (match):   98-136(99)
#03: Initial lookup (miss):   73-102(75)
#04: Initial lookup (random):   230-282(233)
#05: Initial delete all:   66-102(69)
#06: Initial re-inserting:   62-99(64)
#07: Deleting first half:   43-52(43)
#08: Adding (a different) half:   101-156(106)
#09: Lookup after half-change (match):   114-156(120)
#10: Lookup after half-change (miss):   94-103(95)
#11: Churn 1:   98-105(99)
#12: Churn 2:   96-104(98)
#13: Churn 3:   174-184(176)
#14: Post-Churn lookup (match):   93-112(94)
#15: Post-Churn lookup (miss):   77-107(79)
#16: Post-Churn lookup (random):   229-265(232)

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
(Imported from CCAN commit 5c559e7df1d31b4c0ddf26451fac972dc8a0c2c9)

2 files changed, 260 insertions, 1 deletions

diff --git a/lib/ccan/htable/tools/Makefile b/lib/ccan/htable/tools/Makefile
index 001e160b78..289d92b335 100644
--- a/lib/ccan/htable/tools/Makefile
+++ b/lib/ccan/htable/tools/Makefile
@@ -1,5 +1,18 @@
 CFLAGS=-Wall -Werror -O3 -I../../..
+#CFLAGS=-Wall -Werror -g -I../../..
 
-speed: speed.o ../../hash.o
+all: speed stringspeed
+
+speed: speed.o hash.o
 
 speed.o: speed.c ../htable.h ../htable.c
+
+hash.o: ../../hash/hash.c
+	$(CC) $(CFLAGS) -c -o $@ $<
+
+stringspeed: stringspeed.o hash.o ../../talloc.o ../../str_talloc.o ../../grab_file.o ../../str.o ../../time.o ../../noerr.o
+
+stringspeed.o: speed.c ../htable.h ../htable.c
+
+clean:
+	rm -f stringspeed speed
diff --git a/lib/ccan/htable/tools/stringspeed.c b/lib/ccan/htable/tools/stringspeed.c
new file mode 100644
index 0000000000..00ad2790cd
--- /dev/null
+++ b/lib/ccan/htable/tools/stringspeed.c
@@ -0,0 +1,246 @@
+/* Simple speed tests for a hash of strings. */
+#include <ccan/htable/htable_type.h>
+#include <ccan/htable/htable.c>
+#include <ccan/str_talloc/str_talloc.h>
+#include <ccan/grab_file/grab_file.h>
+#include <ccan/talloc/talloc.h>
+#include <ccan/hash/hash.h>
+#include <ccan/time/time.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <time.h>
+#include <unistd.h>
+#include <sys/time.h>
+
+static size_t hashcount;
+
+static const char *strkey(const char *str)
+{
+	return str;
+}
+
+static size_t hash_str(const char *key)
+{
+	hashcount++;
+	return hash(key, strlen(key), 0);
+}
+
+static bool cmp(const char *obj, const char *key)
+{
+	return strcmp(obj, key) == 0;
+}
+
+HTABLE_DEFINE_TYPE(char, strkey, hash_str, cmp, str);
+
+/* Nanoseconds per operation */
+static size_t normalize(const struct timeval *start,
+			const struct timeval *stop,
+			unsigned int num)
+{
+	struct timeval diff;
+
+	timersub(stop, start, &diff);
+
+	/* Floating point is more accurate here. */
+	return (double)(diff.tv_sec * 1000000 + diff.tv_usec)
+		/ num * 1000;
+}
+
+int main(int argc, char *argv[])
+{
+	size_t i, j, num;
+	struct timeval start, stop;
+	struct htable_str *ht;
+	char **words, **misswords;
+
+	words = strsplit(NULL, grab_file(NULL,
+					 argv[1] ? argv[1] : "/usr/share/dict/words",
+					 NULL), "\n");
+	ht = htable_str_new();
+	num = talloc_array_length(words) - 1;
+	printf("%zu words\n", num);
+
+	/* Append and prepend last char for miss testing. */
+	misswords = talloc_array(words, char *, num);
+	for (i = 0; i < num; i++) {
+		char lastc;
+		if (strlen(words[i]))
+			lastc = words[i][strlen(words[i])-1];
+		else
+			lastc = 'z';
+		misswords[i] = talloc_asprintf(misswords, "%c%s%c%c",
+					       lastc, words[i], lastc, lastc);
+	}
+
+	printf("#01: Initial insert: ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 0; i < num; i++)
+		htable_str_add(ht, words[i]);
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	printf("Bytes allocated: %zu\n",
+	       sizeof(((struct htable *)ht)->table[0])
+	       << ((struct htable *)ht)->bits);
+
+	printf("#02: Initial lookup (match): ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 0; i < num; i++)
+		if (htable_str_get(ht, words[i]) != words[i])
+			abort();
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	printf("#03: Initial lookup (miss): ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 0; i < num; i++) {
+		if (htable_str_get(ht, misswords[i]))
+			abort();
+	}
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	/* Lookups in order are very cache-friendly for judy; try random */
+	printf("#04: Initial lookup (random): ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 0, j = 0; i < num; i++, j = (j + 10007) % num)
+		if (htable_str_get(ht, words[j]) != words[j])
+			abort();
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	hashcount = 0;
+	printf("#05: Initial delete all: ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 0; i < num; i++)
+		if (!htable_str_del(ht, words[i]))
+			abort();
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	printf("#06: Initial re-inserting: ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 0; i < num; i++)
+		htable_str_add(ht, words[i]);
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	hashcount = 0;
+	printf("#07: Deleting first half: ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 0; i < num; i+=2)
+		if (!htable_str_del(ht, words[i]))
+			abort();
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	printf("#08: Adding (a different) half: ");
+	fflush(stdout);
+
+	start = time_now();
+	for (i = 0; i < num; i+=2)
+		htable_str_add(ht, misswords[i]);
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	printf("#09: Lookup after half-change (match): ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 1; i < num; i+=2)
+		if (htable_str_get(ht, words[i]) != words[i])
+			abort();
+	for (i = 0; i < num; i+=2) {
+		if (htable_str_get(ht, misswords[i]) != misswords[i])
+			abort();
+	}
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	printf("#10: Lookup after half-change (miss): ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 0; i < num; i+=2)
+		if (htable_str_get(ht, words[i]))
+			abort();
+	for (i = 1; i < num; i+=2) {
+		if (htable_str_get(ht, misswords[i]))
+			abort();
+	}
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	/* Hashtables with delete markers can fill with markers over time.
+	 * so do some changes to see how it operates in long-term. */
+	printf("#11: Churn 1: ");
+	start = time_now();
+	for (j = 0; j < num; j+=2) {
+		if (!htable_str_del(ht, misswords[j]))
+			abort();
+		if (!htable_str_add(ht, words[j]))
+			abort();
+	}
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	printf("#12: Churn 2: ");
+	start = time_now();
+	for (j = 1; j < num; j+=2) {
+		if (!htable_str_del(ht, words[j]))
+			abort();
+		if (!htable_str_add(ht, misswords[j]))
+			abort();
+	}
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	printf("#13: Churn 3: ");
+	start = time_now();
+	for (j = 1; j < num; j+=2) {
+		if (!htable_str_del(ht, misswords[j]))
+			abort();
+		if (!htable_str_add(ht, words[j]))
+			abort();
+	}
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	/* Now it's back to normal... */
+	printf("#14: Post-Churn lookup (match): ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 0; i < num; i++)
+		if (htable_str_get(ht, words[i]) != words[i])
+			abort();
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	printf("#15: Post-Churn lookup (miss): ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 0; i < num; i++) {
+		if (htable_str_get(ht, misswords[i]))
+			abort();
+	}
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	/* Lookups in order are very cache-friendly for judy; try random */
+	printf("#16: Post-Churn lookup (random): ");
+	fflush(stdout);
+	start = time_now();
+	for (i = 0, j = 0; i < num; i++, j = (j + 10007) % num)
+		if (htable_str_get(ht, words[j]) != words[j])
+			abort();
+	stop = time_now();
+	printf(" %zu ns\n", normalize(&start, &stop, num));
+
+	return 0;
+}