[deliverable/linux.git] / tools / perf / bench / mem-memcpy.c

/*
 * mem-memcpy.c
 *
 * memcpy: Simple memory copy in various ways
 *
 * Written by Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp>
 */

#include "../perf.h"
#include "../util/util.h"
#include "../util/parse-options.h"
#include "../util/header.h"
#include "bench.h"
#include "mem-memcpy-arch.h"

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/time.h>
#include <errno.h>

#define K 1024

static const char	*length_str	= "1MB";
static const char	*routine	= "default";
static int		iterations	= 1;
static bool		use_cycle;
static int		cycle_fd;
static bool		only_prefault;
static bool		no_prefault;

static const struct option options[] = {
	OPT_STRING('l', "length", &length_str, "1MB",
		    "Specify length of memory to copy. "
		    "Available units: B, KB, MB, GB and TB (upper and lower)"),
	OPT_STRING('r', "routine", &routine, "default",
		    "Specify routine to copy"),
	OPT_INTEGER('i', "iterations", &iterations,
		    "repeat memcpy() invocation this number of times"),
	OPT_BOOLEAN('c', "cycle", &use_cycle,
		    "Use cycles event instead of gettimeofday() for measuring"),
	OPT_BOOLEAN('o', "only-prefault", &only_prefault,
		    "Show only the result with page faults before memcpy()"),
	OPT_BOOLEAN('n', "no-prefault", &no_prefault,
		    "Show only the result without page faults before memcpy()"),
	OPT_END()
};

typedef void *(*memcpy_t)(void *, const void *, size_t);

struct routine {
	const char *name;
	const char *desc;
	memcpy_t fn;
};

struct routine routines[] = {
	{ "default",
	  "Default memcpy() provided by glibc",
	  memcpy },
#ifdef HAVE_ARCH_X86_64_SUPPORT

#define MEMCPY_FN(fn, name, desc) { name, desc, fn },
#include "mem-memcpy-x86-64-asm-def.h"
#undef MEMCPY_FN

#endif

	{ NULL,
	  NULL,
	  NULL   }
};

static const char * const bench_mem_memcpy_usage[] = {
	"perf bench mem memcpy <options>",
	NULL
};

static struct perf_event_attr cycle_attr = {
	.type		= PERF_TYPE_HARDWARE,
	.config		= PERF_COUNT_HW_CPU_CYCLES
};

static void init_cycle(void)
{
	cycle_fd = sys_perf_event_open(&cycle_attr, getpid(), -1, -1, 0);

	if (cycle_fd < 0 && errno == ENOSYS)
		die("No CONFIG_PERF_EVENTS=y kernel support configured?\n");
	else
		BUG_ON(cycle_fd < 0);
}

static u64 get_cycle(void)
{
	int ret;
	u64 clk;

	ret = read(cycle_fd, &clk, sizeof(u64));
	BUG_ON(ret != sizeof(u64));

	return clk;
}

static double timeval2double(struct timeval *ts)
{
	return (double)ts->tv_sec +
		(double)ts->tv_usec / (double)1000000;
}

static void alloc_mem(void **dst, void **src, size_t length)
{
	*dst = zalloc(length);
	if (!*dst)
		die("memory allocation failed - maybe length is too large?\n");

	*src = zalloc(length);
	if (!*src)
		die("memory allocation failed - maybe length is too large?\n");
	/* Make sure to always replace the zero pages even if MMAP_THRESH is crossed */
	memset(*src, 0, length);
}

static u64 do_memcpy_cycle(memcpy_t fn, size_t len, bool prefault)
{
	u64 cycle_start = 0ULL, cycle_end = 0ULL;
	void *src = NULL, *dst = NULL;
	int i;

	alloc_mem(&src, &dst, len);

	if (prefault)
		fn(dst, src, len);

	cycle_start = get_cycle();
	for (i = 0; i < iterations; ++i)
		fn(dst, src, len);
	cycle_end = get_cycle();

	free(src);
	free(dst);
	return cycle_end - cycle_start;
}

static double do_memcpy_gettimeofday(memcpy_t fn, size_t len, bool prefault)
{
	struct timeval tv_start, tv_end, tv_diff;
	void *src = NULL, *dst = NULL;
	int i;

	alloc_mem(&src, &dst, len);

	if (prefault)
		fn(dst, src, len);

	BUG_ON(gettimeofday(&tv_start, NULL));
	for (i = 0; i < iterations; ++i)
		fn(dst, src, len);
	BUG_ON(gettimeofday(&tv_end, NULL));

	timersub(&tv_end, &tv_start, &tv_diff);

	free(src);
	free(dst);
	return (double)((double)len / timeval2double(&tv_diff));
}

#define pf (no_prefault ? 0 : 1)

#define print_bps(x) do {					\
		if (x < K)					\
			printf(" %14lf B/Sec", x);		\
		else if (x < K * K)				\
			printf(" %14lfd KB/Sec", x / K);	\
		else if (x < K * K * K)				\
			printf(" %14lf MB/Sec", x / K / K);	\
		else						\
			printf(" %14lf GB/Sec", x / K / K / K); \
	} while (0)

int bench_mem_memcpy(int argc, const char **argv,
		     const char *prefix __maybe_unused)
{
	int i;
	size_t len;
	double result_bps[2];
	u64 result_cycle[2];

	argc = parse_options(argc, argv, options,
			     bench_mem_memcpy_usage, 0);

	if (no_prefault && only_prefault) {
		fprintf(stderr, "Invalid options: -o and -n are mutually exclusive\n");
		return 1;
	}

	if (use_cycle)
		init_cycle();

	len = (size_t)perf_atoll((char *)length_str);

	result_cycle[0] = result_cycle[1] = 0ULL;
	result_bps[0] = result_bps[1] = 0.0;

	if ((s64)len <= 0) {
		fprintf(stderr, "Invalid length:%s\n", length_str);
		return 1;
	}

	/* same to without specifying either of prefault and no-prefault */
	if (only_prefault && no_prefault)
		only_prefault = no_prefault = false;

	for (i = 0; routines[i].name; i++) {
		if (!strcmp(routines[i].name, routine))
			break;
	}
	if (!routines[i].name) {
		printf("Unknown routine:%s\n", routine);
		printf("Available routines...\n");
		for (i = 0; routines[i].name; i++) {
			printf("\t%s ... %s\n",
			       routines[i].name, routines[i].desc);
		}
		return 1;
	}

	if (bench_format == BENCH_FORMAT_DEFAULT)
		printf("# Copying %s Bytes ...\n\n", length_str);

	if (!only_prefault && !no_prefault) {
		/* show both of results */
		if (use_cycle) {
			result_cycle[0] =
				do_memcpy_cycle(routines[i].fn, len, false);
			result_cycle[1] =
				do_memcpy_cycle(routines[i].fn, len, true);
		} else {
			result_bps[0] =
				do_memcpy_gettimeofday(routines[i].fn,
						len, false);
			result_bps[1] =
				do_memcpy_gettimeofday(routines[i].fn,
						len, true);
		}
	} else {
		if (use_cycle) {
			result_cycle[pf] =
				do_memcpy_cycle(routines[i].fn,
						len, only_prefault);
		} else {
			result_bps[pf] =
				do_memcpy_gettimeofday(routines[i].fn,
						len, only_prefault);
		}
	}

	switch (bench_format) {
	case BENCH_FORMAT_DEFAULT:
		if (!only_prefault && !no_prefault) {
			if (use_cycle) {
				printf(" %14lf Cycle/Byte\n",
					(double)result_cycle[0]
					/ (double)len);
				printf(" %14lf Cycle/Byte (with prefault)\n",
					(double)result_cycle[1]
					/ (double)len);
			} else {
				print_bps(result_bps[0]);
				printf("\n");
				print_bps(result_bps[1]);
				printf(" (with prefault)\n");
			}
		} else {
			if (use_cycle) {
				printf(" %14lf Cycle/Byte",
					(double)result_cycle[pf]
					/ (double)len);
			} else
				print_bps(result_bps[pf]);

			printf("%s\n", only_prefault ? " (with prefault)" : "");
		}
		break;
	case BENCH_FORMAT_SIMPLE:
		if (!only_prefault && !no_prefault) {
			if (use_cycle) {
				printf("%lf %lf\n",
					(double)result_cycle[0] / (double)len,
					(double)result_cycle[1] / (double)len);
			} else {
				printf("%lf %lf\n",
					result_bps[0], result_bps[1]);
			}
		} else {
			if (use_cycle) {
				printf("%lf\n", (double)result_cycle[pf]
					/ (double)len);
			} else
				printf("%lf\n", result_bps[pf]);
		}
		break;
	default:
		/* reaching this means there's some disaster: */
		die("unknown format: %d\n", bench_format);
		break;
	}

	return 0;
}
Commit	Line	Data
827f3b49 HM	1	/*
	2	* mem-memcpy.c
	3	*
	4	* memcpy: Simple memory copy in various ways
	5	*
	6	* Written by Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp>
	7	*/
827f3b49 HM	8
	9	#include "../perf.h"
	10	#include "../util/util.h"
	11	#include "../util/parse-options.h"
827f3b49 HM	12	#include "../util/header.h"
827f3b49 HM	13	#include "bench.h"
49ce8fc6	14	#include "mem-memcpy-arch.h"
827f3b49 HM	15
	16	#include <stdio.h>
	17	#include <stdlib.h>
	18	#include <string.h>
	19	#include <sys/time.h>
	20	#include <errno.h>
	21
	22	#define K 1024
	23
12eac0bf HM	24	static const char *length_str = "1MB";
12eac0bf HM	25	static const char *routine = "default";
e3e877e7	26	static int iterations = 1;
17d7a112 HM	27	static bool use_cycle;
17d7a112 HM	28	static int cycle_fd;
49ce8fc6 HM	29	static bool only_prefault;
49ce8fc6 HM	30	static bool no_prefault;
827f3b49 HM	31
	32	static const struct option options[] = {
	33	OPT_STRING('l', "length", &length_str, "1MB",
	34	"Specify length of memory to copy. "
08942f6d	35	"Available units: B, KB, MB, GB and TB (upper and lower)"),
827f3b49 HM	36	OPT_STRING('r', "routine", &routine, "default",
827f3b49 HM	37	"Specify routine to copy"),
e3e877e7 JB	38	OPT_INTEGER('i', "iterations", &iterations,
e3e877e7 JB	39	"repeat memcpy() invocation this number of times"),
17d7a112	40	OPT_BOOLEAN('c', "cycle", &use_cycle,
08942f6d	41	"Use cycles event instead of gettimeofday() for measuring"),
49ce8fc6 HM	42	OPT_BOOLEAN('o', "only-prefault", &only_prefault,
	43	"Show only the result with page faults before memcpy()"),
	44	OPT_BOOLEAN('n', "no-prefault", &no_prefault,
	45	"Show only the result without page faults before memcpy()"),
827f3b49 HM	46	OPT_END()
	47	};
	48
49ce8fc6 HM	49	typedef void (memcpy_t)(void , const void , size_t);
49ce8fc6 HM	50
827f3b49 HM	51	struct routine {
	52	const char *name;
	53	const char *desc;
49ce8fc6	54	memcpy_t fn;
827f3b49 HM	55	};
	56
	57	struct routine routines[] = {
	58	{ "default",
	59	"Default memcpy() provided by glibc",
	60	memcpy },
89fe808a	61	#ifdef HAVE_ARCH_X86_64_SUPPORT
49ce8fc6 HM	62
	63	#define MEMCPY_FN(fn, name, desc) { name, desc, fn },
	64	#include "mem-memcpy-x86-64-asm-def.h"
	65	#undef MEMCPY_FN
	66
	67	#endif
	68
827f3b49 HM	69	{ NULL,
	70	NULL,
	71	NULL }
	72	};
	73
	74	static const char * const bench_mem_memcpy_usage[] = {
	75	"perf bench mem memcpy <options>",
	76	NULL
	77	};
	78
17d7a112	79	static struct perf_event_attr cycle_attr = {
12eac0bf HM	80	.type = PERF_TYPE_HARDWARE,
12eac0bf HM	81	.config = PERF_COUNT_HW_CPU_CYCLES
827f3b49 HM	82	};
827f3b49 HM	83
17d7a112	84	static void init_cycle(void)
827f3b49	85	{
17d7a112	86	cycle_fd = sys_perf_event_open(&cycle_attr, getpid(), -1, -1, 0);
12eac0bf	87
17d7a112	88	if (cycle_fd < 0 && errno == ENOSYS)
12eac0bf HM	89	die("No CONFIG_PERF_EVENTS=y kernel support configured?\n");
12eac0bf HM	90	else
17d7a112	91	BUG_ON(cycle_fd < 0);
827f3b49 HM	92	}
827f3b49 HM	93
17d7a112	94	static u64 get_cycle(void)
827f3b49 HM	95	{
	96	int ret;
	97	u64 clk;
	98
17d7a112	99	ret = read(cycle_fd, &clk, sizeof(u64));
827f3b49 HM	100	BUG_ON(ret != sizeof(u64));
	101
	102	return clk;
	103	}
	104
	105	static double timeval2double(struct timeval *ts)
	106	{
	107	return (double)ts->tv_sec +
	108	(double)ts->tv_usec / (double)1000000;
	109	}
	110
49ce8fc6 HM	111	static void alloc_mem(void dst, void src, size_t length)
	112	{
	113	*dst = zalloc(length);
13966721	114	if (!*dst)
49ce8fc6 HM	115	die("memory allocation failed - maybe length is too large?\n");
	116
	117	*src = zalloc(length);
13966721	118	if (!*src)
49ce8fc6	119	die("memory allocation failed - maybe length is too large?\n");
a198996c AK	120	/* Make sure to always replace the zero pages even if MMAP_THRESH is crossed */
a198996c AK	121	memset(*src, 0, length);
49ce8fc6 HM	122	}
49ce8fc6 HM	123
17d7a112	124	static u64 do_memcpy_cycle(memcpy_t fn, size_t len, bool prefault)
49ce8fc6	125	{
17d7a112	126	u64 cycle_start = 0ULL, cycle_end = 0ULL;
49ce8fc6	127	void src = NULL, dst = NULL;
e3e877e7	128	int i;
49ce8fc6 HM	129
	130	alloc_mem(&src, &dst, len);
	131
	132	if (prefault)
	133	fn(dst, src, len);
	134
17d7a112	135	cycle_start = get_cycle();
e3e877e7 JB	136	for (i = 0; i < iterations; ++i)
e3e877e7 JB	137	fn(dst, src, len);
17d7a112	138	cycle_end = get_cycle();
49ce8fc6 HM	139
	140	free(src);
	141	free(dst);
17d7a112	142	return cycle_end - cycle_start;
49ce8fc6 HM	143	}
	144
	145	static double do_memcpy_gettimeofday(memcpy_t fn, size_t len, bool prefault)
	146	{
	147	struct timeval tv_start, tv_end, tv_diff;
	148	void src = NULL, dst = NULL;
e3e877e7	149	int i;
49ce8fc6 HM	150
	151	alloc_mem(&src, &dst, len);
	152
	153	if (prefault)
	154	fn(dst, src, len);
	155
	156	BUG_ON(gettimeofday(&tv_start, NULL));
e3e877e7 JB	157	for (i = 0; i < iterations; ++i)
e3e877e7 JB	158	fn(dst, src, len);
49ce8fc6 HM	159	BUG_ON(gettimeofday(&tv_end, NULL));
	160
	161	timersub(&tv_end, &tv_start, &tv_diff);
	162
	163	free(src);
	164	free(dst);
	165	return (double)((double)len / timeval2double(&tv_diff));
	166	}
	167
	168	#define pf (no_prefault ? 0 : 1)
	169
	170	#define print_bps(x) do { \
	171	if (x < K) \
	172	printf(" %14lf B/Sec", x); \
	173	else if (x < K * K) \
	174	printf(" %14lfd KB/Sec", x / K); \
	175	else if (x < K * K * K) \
	176	printf(" %14lf MB/Sec", x / K / K); \
	177	else \
	178	printf(" %14lf GB/Sec", x / K / K / K); \
	179	} while (0)
	180
827f3b49	181	int bench_mem_memcpy(int argc, const char **argv,
1d037ca1	182	const char *prefix __maybe_unused)
827f3b49 HM	183	{
827f3b49 HM	184	int i;
49ce8fc6 HM	185	size_t len;
49ce8fc6 HM	186	double result_bps[2];
17d7a112	187	u64 result_cycle[2];
827f3b49	188
827f3b49 HM	189	argc = parse_options(argc, argv, options,
	190	bench_mem_memcpy_usage, 0);
	191
424e9634 DB	192	if (no_prefault && only_prefault) {
	193	fprintf(stderr, "Invalid options: -o and -n are mutually exclusive\n");
	194	return 1;
	195	}
	196
17d7a112 HM	197	if (use_cycle)
17d7a112 HM	198	init_cycle();
49ce8fc6 HM	199
49ce8fc6 HM	200	len = (size_t)perf_atoll((char *)length_str);
12eac0bf	201
17d7a112	202	result_cycle[0] = result_cycle[1] = 0ULL;
49ce8fc6 HM	203	result_bps[0] = result_bps[1] = 0.0;
	204
	205	if ((s64)len <= 0) {
827f3b49 HM	206	fprintf(stderr, "Invalid length:%s\n", length_str);
	207	return 1;
	208	}
	209
49ce8fc6 HM	210	/* same to without specifying either of prefault and no-prefault */
	211	if (only_prefault && no_prefault)
	212	only_prefault = no_prefault = false;
	213
827f3b49 HM	214	for (i = 0; routines[i].name; i++) {
	215	if (!strcmp(routines[i].name, routine))
	216	break;
	217	}
	218	if (!routines[i].name) {
	219	printf("Unknown routine:%s\n", routine);
	220	printf("Available routines...\n");
	221	for (i = 0; routines[i].name; i++) {
	222	printf("\t%s ... %s\n",
	223	routines[i].name, routines[i].desc);
	224	}
	225	return 1;
	226	}
	227
49ce8fc6 HM	228	if (bench_format == BENCH_FORMAT_DEFAULT)
49ce8fc6 HM	229	printf("# Copying %s Bytes ...\n\n", length_str);
827f3b49	230
49ce8fc6 HM	231	if (!only_prefault && !no_prefault) {
49ce8fc6 HM	232	/* show both of results */
17d7a112 HM	233	if (use_cycle) {
	234	result_cycle[0] =
	235	do_memcpy_cycle(routines[i].fn, len, false);
	236	result_cycle[1] =
	237	do_memcpy_cycle(routines[i].fn, len, true);
49ce8fc6 HM	238	} else {
	239	result_bps[0] =
	240	do_memcpy_gettimeofday(routines[i].fn,
	241	len, false);
	242	result_bps[1] =
	243	do_memcpy_gettimeofday(routines[i].fn,
	244	len, true);
	245	}
827f3b49	246	} else {
17d7a112 HM	247	if (use_cycle) {
	248	result_cycle[pf] =
	249	do_memcpy_cycle(routines[i].fn,
49ce8fc6 HM	250	len, only_prefault);
	251	} else {
	252	result_bps[pf] =
	253	do_memcpy_gettimeofday(routines[i].fn,
	254	len, only_prefault);
	255	}
827f3b49 HM	256	}
	257
	258	switch (bench_format) {
	259	case BENCH_FORMAT_DEFAULT:
49ce8fc6	260	if (!only_prefault && !no_prefault) {
17d7a112 HM	261	if (use_cycle) {
	262	printf(" %14lf Cycle/Byte\n",
	263	(double)result_cycle[0]
49ce8fc6	264	/ (double)len);
17d7a112 HM	265	printf(" %14lf Cycle/Byte (with prefault)\n",
17d7a112 HM	266	(double)result_cycle[1]
49ce8fc6 HM	267	/ (double)len);
	268	} else {
	269	print_bps(result_bps[0]);
	270	printf("\n");
	271	print_bps(result_bps[1]);
	272	printf(" (with prefault)\n");
827f3b49	273	}
49ce8fc6	274	} else {
17d7a112 HM	275	if (use_cycle) {
	276	printf(" %14lf Cycle/Byte",
	277	(double)result_cycle[pf]
49ce8fc6 HM	278	/ (double)len);
	279	} else
	280	print_bps(result_bps[pf]);
	281
	282	printf("%s\n", only_prefault ? " (with prefault)" : "");
827f3b49 HM	283	}
	284	break;
	285	case BENCH_FORMAT_SIMPLE:
49ce8fc6	286	if (!only_prefault && !no_prefault) {
17d7a112	287	if (use_cycle) {
49ce8fc6	288	printf("%lf %lf\n",
17d7a112 HM	289	(double)result_cycle[0] / (double)len,
17d7a112 HM	290	(double)result_cycle[1] / (double)len);
49ce8fc6 HM	291	} else {
	292	printf("%lf %lf\n",
	293	result_bps[0], result_bps[1]);
	294	}
	295	} else {
17d7a112 HM	296	if (use_cycle) {
17d7a112 HM	297	printf("%lf\n", (double)result_cycle[pf]
49ce8fc6 HM	298	/ (double)len);
	299	} else
	300	printf("%lf\n", result_bps[pf]);
	301	}
827f3b49 HM	302	break;
827f3b49 HM	303	default:
12eac0bf HM	304	/* reaching this means there's some disaster: */
12eac0bf HM	305	die("unknown format: %d\n", bench_format);
827f3b49 HM	306	break;
	307	}
	308
	309	return 0;
	310	}