From 054b3b48c9ccfde24f4c01fb6f59773cfcd422d3 Mon Sep 17 00:00:00 2001
From: Paweł Redman <pawel.redman@gmail.com>
Date: Wed, 16 May 2018 19:33:28 +0200
Subject: Refactor statistics.

---
 Makefile  |  2 +-
 turbowc.c | 64 +++++++++++++++++++++++++++++++++++++++++++--------------------
 2 files changed, 45 insertions(+), 21 deletions(-)

diff --git a/Makefile b/Makefile
index 93a7aad..39b9324 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,5 @@
 CC       ?= cc
-CFLAGS   += -O3 -march=native
+CFLAGS   += -O3 -march=native -Wall
 CPPFLAGS += 
 LDFLAGS  += -pthread
 
diff --git a/turbowc.c b/turbowc.c
index c556ed5..16512f6 100644
--- a/turbowc.c
+++ b/turbowc.c
@@ -11,6 +11,15 @@
 #include <sys/mman.h>
 #include <time.h>
 
+typedef uint64_t ntime_t;
+
+ntime_t nclock(void)
+{
+	struct timespec ts;
+	clock_gettime(CLOCK_MONOTONIC_RAW, &ts);
+	return (ntime_t)ts.tv_sec * 1000000000 + ts.tv_nsec;
+}
+
 static size_t count_newlines(void *data, size_t size)
 {
 	size_t i, total = 0;
@@ -103,15 +112,18 @@ static void *thread_f(void *arg)
 	return NULL;
 }
 
+struct {
+	size_t size, lines;
+	ntime_t time;
+} totals;
+
 static int turbowc(const char *path, size_t num_threads)
 {
-	int ret = 1, opt, fd;
+	int ret = 1, fd;
 	struct stat stat;
 	void *map;
 	size_t i;
-	pthread_t *threads;
-	struct timespec t0, t1;
-	double delta;
+	pthread_t *threads = NULL;
 
 	fd = open(path, O_RDONLY);
 	if (fd == -1) {
@@ -133,6 +145,7 @@ static int turbowc(const char *path, size_t num_threads)
 
 	data_pool.cursor = map;
 	data_pool.left = stat.st_size;
+	data_pool.total = 0;
 
 	if (num_threads) {
 		threads = calloc(num_threads, sizeof(pthread_t));
@@ -142,8 +155,6 @@ static int turbowc(const char *path, size_t num_threads)
 		}
 	}
 
-	clock_gettime(CLOCK_MONOTONIC, &t0);
-
 	for (i = 0; i < num_threads; i++) {
 		int rv;
 
@@ -164,20 +175,10 @@ static int turbowc(const char *path, size_t num_threads)
 		pthread_join(threads[i], NULL);
 
 	data_pool.total += count_newlines(data_pool.cursor, data_pool.left);
-	clock_gettime(CLOCK_MONOTONIC, &t1);
-
 	printf("%zu %s\n", data_pool.total, path);
 
-	delta = (t1.tv_sec - t0.tv_sec) +
-		((double)t1.tv_nsec - t0.tv_nsec) * 1.0e-9;
-
-	fprintf(stderr, " Total size: %zu\n", stat.st_size);
-	fprintf(stderr, " Total time: %f s\n", delta);
-	fprintf(stderr, "Time / char: %f ns\n", delta / stat.st_size * 1.0e+9);
-	fprintf(stderr, " Throughput: %f GB/s\n", stat.st_size / delta * 1.0e-9);
-	fprintf(stderr, "Total lines: %zu\n", data_pool.total);
-	fprintf(stderr, "  Avg. line: %f char(s)\n", (double)stat.st_size / data_pool.total);
-	fprintf(stderr, "Time / line: %f ns\n", delta / data_pool.total * 1.0e+9);
+	totals.size += stat.st_size;
+	totals.lines += data_pool.total;
 
 	ret = 0;
 
@@ -197,8 +198,10 @@ int main(int argc, char **argv)
 {
 	int opt;
 	size_t num_threads = 1;
+	_Bool print_stats = 0;
+	ntime_t t0;
 
-	while ((opt = getopt(argc, argv, "vt:")) != -1) {
+	while ((opt = getopt(argc, argv, "vt:s")) != -1) {
 		switch (opt) {
 		case 'v':
 			printf("turbowc, with mmap, SSE2 and multithreading\n");
@@ -208,9 +211,13 @@ int main(int argc, char **argv)
 			num_threads = strtoull(optarg, NULL, 10);
 			break;
 
+		case 's':
+			print_stats = 1;
+			break;
+
 		default:
 		print_usage:
-			fprintf(stderr, "usage: turbowc [-v] [-t THREADS] [FILE]\n");
+			fprintf(stderr, "usage: turbowc [-v] [-s] [-t THREADS] [FILE]\n");
 			return 1;
 		}
 	}
@@ -218,9 +225,26 @@ int main(int argc, char **argv)
 	if (optind >= argc)
 		goto print_usage;
 
+	t0 = nclock();
+
 	for (; optind < argc; optind++ )
 		if (turbowc(argv[optind], num_threads))
 			return 1;
 
+	totals.time = nclock() - t0;
+
+	if (!print_stats)
+		goto skip_stats;
+	printf("Lines\t%zu\n", totals.lines);
+	printf("Size\t%zu\t%f\t-\n",
+		totals.size,
+		(double)totals.size / totals.lines);
+	printf("Time\t%f s\t%f ns\t%f ps\n",
+		totals.time * 1.0e-9,
+		(double)totals.time / totals.lines,
+		totals.time * 1.0e+3 / totals.size);
+	printf("Thput\t%f GB/s\n", (double)totals.size / totals.time);
+
+skip_stats:
 	return 0;
 }
-- 
cgit