* [PATCH 1/3] perf: bench: prepare memcpy for merge
@ 2014-12-02 15:50 Rabin Vincent
2014-12-02 15:50 ` [PATCH 2/3] perf: bench: merge memset into memcpy Rabin Vincent
` (2 more replies)
0 siblings, 3 replies; 7+ messages in thread
From: Rabin Vincent @ 2014-12-02 15:50 UTC (permalink / raw)
To: Peter Zijlstra, Paul Mackerras, Ingo Molnar,
Arnaldo Carvalho de Melo
Cc: linux-kernel, rabin, Rabin Vincent
The memset benchmark is largely copy-pasted from the memcpy benchmark.
Prepare the memcpy file for merge with memset by extracting out a
generic function.
Signed-off-by: Rabin Vincent <rabin.vincent@axis.com>
---
tools/perf/bench/mem-memcpy.c | 182 +++++++++++++++++++++++------------------
1 file changed, 104 insertions(+), 78 deletions(-)
diff --git a/tools/perf/bench/mem-memcpy.c b/tools/perf/bench/mem-memcpy.c
index 2465141..4a4493a 100644
--- a/tools/perf/bench/mem-memcpy.c
+++ b/tools/perf/bench/mem-memcpy.c
@@ -52,16 +52,18 @@ typedef void *(*memcpy_t)(void *, const void *, size_t);
struct routine {
const char *name;
const char *desc;
- memcpy_t fn;
+ union {
+ memcpy_t memcpy;
+ } fn;
};
-struct routine routines[] = {
- { "default",
- "Default memcpy() provided by glibc",
- memcpy },
+struct routine memcpy_routines[] = {
+ { .name = "default",
+ .desc = "Default memcpy() provided by glibc",
+ .fn.memcpy = memcpy },
#ifdef HAVE_ARCH_X86_64_SUPPORT
-#define MEMCPY_FN(fn, name, desc) { name, desc, fn },
+#define MEMCPY_FN(_fn, _name, _desc) {.name = _name, .desc = _desc, .fn.memcpy = _fn},
#include "mem-memcpy-x86-64-asm-def.h"
#undef MEMCPY_FN
@@ -69,7 +71,7 @@ struct routine routines[] = {
{ NULL,
NULL,
- NULL }
+ {NULL} }
};
static const char * const bench_mem_memcpy_usage[] = {
@@ -110,63 +112,6 @@ static double timeval2double(struct timeval *ts)
(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 { \
@@ -180,8 +125,16 @@ static double do_memcpy_gettimeofday(memcpy_t fn, size_t len, bool prefault)
printf(" %14lf GB/Sec", x / K / K / K); \
} while (0)
-int bench_mem_memcpy(int argc, const char **argv,
- const char *prefix __maybe_unused)
+struct bench_mem_info {
+ const struct routine *routines;
+ u64 (*do_cycle)(const struct routine *r, size_t len, bool prefault);
+ double (*do_gettimeofday)(const struct routine *r, size_t len, bool prefault);
+ const char *const *usage;
+};
+
+static int bench_mem_common(int argc, const char **argv,
+ const char *prefix __maybe_unused,
+ struct bench_mem_info *info)
{
int i;
size_t len;
@@ -189,7 +142,7 @@ int bench_mem_memcpy(int argc, const char **argv,
u64 result_cycle[2];
argc = parse_options(argc, argv, options,
- bench_mem_memcpy_usage, 0);
+ info->usage, 0);
if (no_prefault && only_prefault) {
fprintf(stderr, "Invalid options: -o and -n are mutually exclusive\n");
@@ -213,16 +166,16 @@ int bench_mem_memcpy(int argc, const char **argv,
if (only_prefault && no_prefault)
only_prefault = no_prefault = false;
- for (i = 0; routines[i].name; i++) {
- if (!strcmp(routines[i].name, routine))
+ for (i = 0; info->routines[i].name; i++) {
+ if (!strcmp(info->routines[i].name, routine))
break;
}
- if (!routines[i].name) {
+ if (!info->routines[i].name) {
printf("Unknown routine:%s\n", routine);
printf("Available routines...\n");
- for (i = 0; routines[i].name; i++) {
+ for (i = 0; info->routines[i].name; i++) {
printf("\t%s ... %s\n",
- routines[i].name, routines[i].desc);
+ info->routines[i].name, info->routines[i].desc);
}
return 1;
}
@@ -234,25 +187,25 @@ int bench_mem_memcpy(int argc, const char **argv,
/* show both of results */
if (use_cycle) {
result_cycle[0] =
- do_memcpy_cycle(routines[i].fn, len, false);
+ info->do_cycle(&info->routines[i], len, false);
result_cycle[1] =
- do_memcpy_cycle(routines[i].fn, len, true);
+ info->do_cycle(&info->routines[i], len, true);
} else {
result_bps[0] =
- do_memcpy_gettimeofday(routines[i].fn,
+ info->do_gettimeofday(&info->routines[i],
len, false);
result_bps[1] =
- do_memcpy_gettimeofday(routines[i].fn,
+ info->do_gettimeofday(&info->routines[i],
len, true);
}
} else {
if (use_cycle) {
result_cycle[pf] =
- do_memcpy_cycle(routines[i].fn,
+ info->do_cycle(&info->routines[i],
len, only_prefault);
} else {
result_bps[pf] =
- do_memcpy_gettimeofday(routines[i].fn,
+ info->do_gettimeofday(&info->routines[i],
len, only_prefault);
}
}
@@ -310,3 +263,76 @@ int bench_mem_memcpy(int argc, const char **argv,
return 0;
}
+
+static void memcpy_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(const struct routine *r, size_t len, bool prefault)
+{
+ u64 cycle_start = 0ULL, cycle_end = 0ULL;
+ void *src = NULL, *dst = NULL;
+ memcpy_t fn = r->fn.memcpy;
+ int i;
+
+ memcpy_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(const struct routine *r, size_t len,
+ bool prefault)
+{
+ struct timeval tv_start, tv_end, tv_diff;
+ memcpy_t fn = r->fn.memcpy;
+ void *src = NULL, *dst = NULL;
+ int i;
+
+ memcpy_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));
+}
+
+int bench_mem_memcpy(int argc, const char **argv,
+ const char *prefix __maybe_unused)
+{
+ struct bench_mem_info info = {
+ .routines = memcpy_routines,
+ .do_cycle = do_memcpy_cycle,
+ .do_gettimeofday = do_memcpy_gettimeofday,
+ .usage = bench_mem_memcpy_usage,
+ };
+
+ return bench_mem_common(argc, argv, prefix, &info);
+}
--
1.7.10.4
^ permalink raw reply related [flat|nested] 7+ messages in thread
* [PATCH 2/3] perf: bench: merge memset into memcpy
2014-12-02 15:50 [PATCH 1/3] perf: bench: prepare memcpy for merge Rabin Vincent
@ 2014-12-02 15:50 ` Rabin Vincent
2014-12-12 8:16 ` [tip:perf/urgent] perf bench: Merge " tip-bot for Rabin Vincent
2014-12-02 15:50 ` [PATCH 3/3] perf: bench: fix memcpy/memset output Rabin Vincent
2014-12-12 8:15 ` [tip:perf/urgent] perf bench: Prepare memcpy for merge tip-bot for Rabin Vincent
2 siblings, 1 reply; 7+ messages in thread
From: Rabin Vincent @ 2014-12-02 15:50 UTC (permalink / raw)
To: Peter Zijlstra, Paul Mackerras, Ingo Molnar,
Arnaldo Carvalho de Melo
Cc: linux-kernel, rabin, Rabin Vincent
The memset benchmark is largely copy-pasted from the memcpy benchmark.
Merge the two now that memcpy is made more generic.
Signed-off-by: Rabin Vincent <rabin.vincent@axis.com>
---
tools/perf/Makefile.perf | 1 -
tools/perf/bench/mem-memcpy.c | 90 ++++++++++++
tools/perf/bench/mem-memset.c | 304 -----------------------------------------
3 files changed, 90 insertions(+), 305 deletions(-)
delete mode 100644 tools/perf/bench/mem-memset.c
diff --git a/tools/perf/Makefile.perf b/tools/perf/Makefile.perf
index 478efa9..763e68f 100644
--- a/tools/perf/Makefile.perf
+++ b/tools/perf/Makefile.perf
@@ -458,7 +458,6 @@ BUILTIN_OBJS += $(OUTPUT)bench/mem-memcpy-x86-64-asm.o
BUILTIN_OBJS += $(OUTPUT)bench/mem-memset-x86-64-asm.o
endif
BUILTIN_OBJS += $(OUTPUT)bench/mem-memcpy.o
-BUILTIN_OBJS += $(OUTPUT)bench/mem-memset.o
BUILTIN_OBJS += $(OUTPUT)bench/futex-hash.o
BUILTIN_OBJS += $(OUTPUT)bench/futex-wake.o
BUILTIN_OBJS += $(OUTPUT)bench/futex-requeue.o
diff --git a/tools/perf/bench/mem-memcpy.c b/tools/perf/bench/mem-memcpy.c
index 4a4493a..e18be70 100644
--- a/tools/perf/bench/mem-memcpy.c
+++ b/tools/perf/bench/mem-memcpy.c
@@ -13,6 +13,7 @@
#include "../util/cloexec.h"
#include "bench.h"
#include "mem-memcpy-arch.h"
+#include "mem-memset-arch.h"
#include <stdio.h>
#include <stdlib.h>
@@ -48,12 +49,14 @@ static const struct option options[] = {
};
typedef void *(*memcpy_t)(void *, const void *, size_t);
+typedef void *(*memset_t)(void *, int, size_t);
struct routine {
const char *name;
const char *desc;
union {
memcpy_t memcpy;
+ memset_t memset;
} fn;
};
@@ -336,3 +339,90 @@ int bench_mem_memcpy(int argc, const char **argv,
return bench_mem_common(argc, argv, prefix, &info);
}
+
+static void memset_alloc_mem(void **dst, size_t length)
+{
+ *dst = zalloc(length);
+ if (!*dst)
+ die("memory allocation failed - maybe length is too large?\n");
+}
+
+static u64 do_memset_cycle(const struct routine *r, size_t len, bool prefault)
+{
+ u64 cycle_start = 0ULL, cycle_end = 0ULL;
+ memset_t fn = r->fn.memset;
+ void *dst = NULL;
+ int i;
+
+ memset_alloc_mem(&dst, len);
+
+ if (prefault)
+ fn(dst, -1, len);
+
+ cycle_start = get_cycle();
+ for (i = 0; i < iterations; ++i)
+ fn(dst, i, len);
+ cycle_end = get_cycle();
+
+ free(dst);
+ return cycle_end - cycle_start;
+}
+
+static double do_memset_gettimeofday(const struct routine *r, size_t len,
+ bool prefault)
+{
+ struct timeval tv_start, tv_end, tv_diff;
+ memset_t fn = r->fn.memset;
+ void *dst = NULL;
+ int i;
+
+ memset_alloc_mem(&dst, len);
+
+ if (prefault)
+ fn(dst, -1, len);
+
+ BUG_ON(gettimeofday(&tv_start, NULL));
+ for (i = 0; i < iterations; ++i)
+ fn(dst, i, len);
+ BUG_ON(gettimeofday(&tv_end, NULL));
+
+ timersub(&tv_end, &tv_start, &tv_diff);
+
+ free(dst);
+ return (double)((double)len / timeval2double(&tv_diff));
+}
+
+static const char * const bench_mem_memset_usage[] = {
+ "perf bench mem memset <options>",
+ NULL
+};
+
+static const struct routine memset_routines[] = {
+ { .name ="default",
+ .desc = "Default memset() provided by glibc",
+ .fn.memset = memset },
+#ifdef HAVE_ARCH_X86_64_SUPPORT
+
+#define MEMSET_FN(_fn, _name, _desc) { .name = _name, .desc = _desc, .fn.memset = _fn },
+#include "mem-memset-x86-64-asm-def.h"
+#undef MEMSET_FN
+
+#endif
+
+ { .name = NULL,
+ .desc = NULL,
+ .fn.memset = NULL }
+};
+
+int bench_mem_memset(int argc, const char **argv,
+ const char *prefix __maybe_unused)
+{
+ struct bench_mem_info info = {
+ .routines = memset_routines,
+ .do_cycle = do_memset_cycle,
+ .do_gettimeofday = do_memset_gettimeofday,
+ .usage = bench_mem_memset_usage,
+ };
+
+ return bench_mem_common(argc, argv, prefix, &info);
+}
diff --git a/tools/perf/bench/mem-memset.c b/tools/perf/bench/mem-memset.c
deleted file mode 100644
index 75fc3e6..0000000
--- a/tools/perf/bench/mem-memset.c
+++ /dev/null
@@ -1,304 +0,0 @@
-/*
- * mem-memset.c
- *
- * memset: Simple memory set in various ways
- *
- * Trivial clone of mem-memcpy.c.
- */
-
-#include "../perf.h"
-#include "../util/util.h"
-#include "../util/parse-options.h"
-#include "../util/header.h"
-#include "../util/cloexec.h"
-#include "bench.h"
-#include "mem-memset-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 set. "
- "Available units: B, KB, MB, GB and TB (upper and lower)"),
- OPT_STRING('r', "routine", &routine, "default",
- "Specify routine to set"),
- OPT_INTEGER('i', "iterations", &iterations,
- "repeat memset() 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 memset()"),
- OPT_BOOLEAN('n', "no-prefault", &no_prefault,
- "Show only the result without page faults before memset()"),
- OPT_END()
-};
-
-typedef void *(*memset_t)(void *, int, size_t);
-
-struct routine {
- const char *name;
- const char *desc;
- memset_t fn;
-};
-
-static const struct routine routines[] = {
- { "default",
- "Default memset() provided by glibc",
- memset },
-#ifdef HAVE_ARCH_X86_64_SUPPORT
-
-#define MEMSET_FN(fn, name, desc) { name, desc, fn },
-#include "mem-memset-x86-64-asm-def.h"
-#undef MEMSET_FN
-
-#endif
-
- { NULL,
- NULL,
- NULL }
-};
-
-static const char * const bench_mem_memset_usage[] = {
- "perf bench mem memset <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,
- perf_event_open_cloexec_flag());
-
- 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, size_t length)
-{
- *dst = zalloc(length);
- if (!*dst)
- die("memory allocation failed - maybe length is too large?\n");
-}
-
-static u64 do_memset_cycle(memset_t fn, size_t len, bool prefault)
-{
- u64 cycle_start = 0ULL, cycle_end = 0ULL;
- void *dst = NULL;
- int i;
-
- alloc_mem(&dst, len);
-
- if (prefault)
- fn(dst, -1, len);
-
- cycle_start = get_cycle();
- for (i = 0; i < iterations; ++i)
- fn(dst, i, len);
- cycle_end = get_cycle();
-
- free(dst);
- return cycle_end - cycle_start;
-}
-
-static double do_memset_gettimeofday(memset_t fn, size_t len, bool prefault)
-{
- struct timeval tv_start, tv_end, tv_diff;
- void *dst = NULL;
- int i;
-
- alloc_mem(&dst, len);
-
- if (prefault)
- fn(dst, -1, len);
-
- BUG_ON(gettimeofday(&tv_start, NULL));
- for (i = 0; i < iterations; ++i)
- fn(dst, i, len);
- BUG_ON(gettimeofday(&tv_end, NULL));
-
- timersub(&tv_end, &tv_start, &tv_diff);
-
- 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_memset(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_memset_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_memset_cycle(routines[i].fn, len, false);
- result_cycle[1] =
- do_memset_cycle(routines[i].fn, len, true);
- } else {
- result_bps[0] =
- do_memset_gettimeofday(routines[i].fn,
- len, false);
- result_bps[1] =
- do_memset_gettimeofday(routines[i].fn,
- len, true);
- }
- } else {
- if (use_cycle) {
- result_cycle[pf] =
- do_memset_cycle(routines[i].fn,
- len, only_prefault);
- } else {
- result_bps[pf] =
- do_memset_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;
-}
--
1.7.10.4
^ permalink raw reply related [flat|nested] 7+ messages in thread
* [PATCH 3/3] perf: bench: fix memcpy/memset output
2014-12-02 15:50 [PATCH 1/3] perf: bench: prepare memcpy for merge Rabin Vincent
2014-12-02 15:50 ` [PATCH 2/3] perf: bench: merge memset into memcpy Rabin Vincent
@ 2014-12-02 15:50 ` Rabin Vincent
2014-12-08 10:25 ` Ingo Molnar
2014-12-12 8:16 ` [tip:perf/urgent] perf bench: Fix " tip-bot for Rabin Vincent
2014-12-12 8:15 ` [tip:perf/urgent] perf bench: Prepare memcpy for merge tip-bot for Rabin Vincent
2 siblings, 2 replies; 7+ messages in thread
From: Rabin Vincent @ 2014-12-02 15:50 UTC (permalink / raw)
To: Peter Zijlstra, Paul Mackerras, Ingo Molnar,
Arnaldo Carvalho de Melo
Cc: linux-kernel, rabin, Rabin Vincent
The memcpy and memset benchmarks return bogus results when iterations >
0 because the iterations value is not taken into account when
calculating the final result:
$ perf bench mem memset --only-prefault --length 1GB --iterations 1
# Running 'mem/memset' benchmark:
# Copying 1GB Bytes ...
20.798669 GB/Sec (with prefault)
$ perf bench mem memset --only-prefault --length 1GB --iterations 10
# Running 'mem/memset' benchmark:
# Copying 1GB Bytes ...
2.086576 GB/Sec (with prefault)
$ perf bench mem memset --only-prefault --length 1GB --iterations 100
# Running 'mem/memset' benchmark:
# Copying 1GB Bytes ...
212.840917 MB/Sec (with prefault)
Fix this.
Signed-off-by: Rabin Vincent <rabin.vincent@axis.com>
---
tools/perf/bench/mem-memcpy.c | 18 ++++++++++--------
1 file changed, 10 insertions(+), 8 deletions(-)
diff --git a/tools/perf/bench/mem-memcpy.c b/tools/perf/bench/mem-memcpy.c
index e18be70..6c14afe 100644
--- a/tools/perf/bench/mem-memcpy.c
+++ b/tools/perf/bench/mem-memcpy.c
@@ -141,6 +141,7 @@ static int bench_mem_common(int argc, const char **argv,
{
int i;
size_t len;
+ double totallen;
double result_bps[2];
u64 result_cycle[2];
@@ -156,6 +157,7 @@ static int bench_mem_common(int argc, const char **argv,
init_cycle();
len = (size_t)perf_atoll((char *)length_str);
+ totallen = (double)len * iterations;
result_cycle[0] = result_cycle[1] = 0ULL;
result_bps[0] = result_bps[1] = 0.0;
@@ -219,10 +221,10 @@ static int bench_mem_common(int argc, const char **argv,
if (use_cycle) {
printf(" %14lf Cycle/Byte\n",
(double)result_cycle[0]
- / (double)len);
+ / totallen);
printf(" %14lf Cycle/Byte (with prefault)\n",
(double)result_cycle[1]
- / (double)len);
+ / totallen);
} else {
print_bps(result_bps[0]);
printf("\n");
@@ -233,7 +235,7 @@ static int bench_mem_common(int argc, const char **argv,
if (use_cycle) {
printf(" %14lf Cycle/Byte",
(double)result_cycle[pf]
- / (double)len);
+ / totallen);
} else
print_bps(result_bps[pf]);
@@ -244,8 +246,8 @@ static int bench_mem_common(int argc, const char **argv,
if (!only_prefault && !no_prefault) {
if (use_cycle) {
printf("%lf %lf\n",
- (double)result_cycle[0] / (double)len,
- (double)result_cycle[1] / (double)len);
+ (double)result_cycle[0] / totallen,
+ (double)result_cycle[1] / totallen);
} else {
printf("%lf %lf\n",
result_bps[0], result_bps[1]);
@@ -253,7 +255,7 @@ static int bench_mem_common(int argc, const char **argv,
} else {
if (use_cycle) {
printf("%lf\n", (double)result_cycle[pf]
- / (double)len);
+ / totallen);
} else
printf("%lf\n", result_bps[pf]);
}
@@ -324,7 +326,7 @@ static double do_memcpy_gettimeofday(const struct routine *r, size_t len,
free(src);
free(dst);
- return (double)((double)len / timeval2double(&tv_diff));
+ return (double)(((double)len * iterations) / timeval2double(&tv_diff));
}
int bench_mem_memcpy(int argc, const char **argv,
@@ -389,7 +391,7 @@ static double do_memset_gettimeofday(const struct routine *r, size_t len,
timersub(&tv_end, &tv_start, &tv_diff);
free(dst);
- return (double)((double)len / timeval2double(&tv_diff));
+ return (double)(((double)len * iterations) / timeval2double(&tv_diff));
}
static const char * const bench_mem_memset_usage[] = {
--
1.7.10.4
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH 3/3] perf: bench: fix memcpy/memset output
2014-12-02 15:50 ` [PATCH 3/3] perf: bench: fix memcpy/memset output Rabin Vincent
@ 2014-12-08 10:25 ` Ingo Molnar
2014-12-12 8:16 ` [tip:perf/urgent] perf bench: Fix " tip-bot for Rabin Vincent
1 sibling, 0 replies; 7+ messages in thread
From: Ingo Molnar @ 2014-12-08 10:25 UTC (permalink / raw)
To: Rabin Vincent, Arnaldo Carvalho de Melo
Cc: Peter Zijlstra, Paul Mackerras, Ingo Molnar,
Arnaldo Carvalho de Melo, linux-kernel, rabin, Rabin Vincent
* Rabin Vincent <rabin.vincent@axis.com> wrote:
> The memcpy and memset benchmarks return bogus results when iterations >
> 0 because the iterations value is not taken into account when
> calculating the final result:
>
> $ perf bench mem memset --only-prefault --length 1GB --iterations 1
> # Running 'mem/memset' benchmark:
> # Copying 1GB Bytes ...
>
> 20.798669 GB/Sec (with prefault)
> $ perf bench mem memset --only-prefault --length 1GB --iterations 10
> # Running 'mem/memset' benchmark:
> # Copying 1GB Bytes ...
>
> 2.086576 GB/Sec (with prefault)
> $ perf bench mem memset --only-prefault --length 1GB --iterations 100
> # Running 'mem/memset' benchmark:
> # Copying 1GB Bytes ...
>
> 212.840917 MB/Sec (with prefault)
>
> Fix this.
>
> Signed-off-by: Rabin Vincent <rabin.vincent@axis.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Thanks,
Ingo
^ permalink raw reply [flat|nested] 7+ messages in thread
* [tip:perf/urgent] perf bench: Prepare memcpy for merge
2014-12-02 15:50 [PATCH 1/3] perf: bench: prepare memcpy for merge Rabin Vincent
2014-12-02 15:50 ` [PATCH 2/3] perf: bench: merge memset into memcpy Rabin Vincent
2014-12-02 15:50 ` [PATCH 3/3] perf: bench: fix memcpy/memset output Rabin Vincent
@ 2014-12-12 8:15 ` tip-bot for Rabin Vincent
2 siblings, 0 replies; 7+ messages in thread
From: tip-bot for Rabin Vincent @ 2014-12-12 8:15 UTC (permalink / raw)
To: linux-tip-commits
Cc: rabinv, hpa, linux-kernel, rabin.vincent, tglx, mingo, mingo,
acme, paulus, a.p.zijlstra
Commit-ID: 308197b9474bcde2cafba2cd19bef46e0c0428bd
Gitweb: http://git.kernel.org/tip/308197b9474bcde2cafba2cd19bef46e0c0428bd
Author: Rabin Vincent <rabin.vincent@axis.com>
AuthorDate: Tue, 2 Dec 2014 16:50:39 +0100
Committer: Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Tue, 9 Dec 2014 09:14:00 -0300
perf bench: Prepare memcpy for merge
The memset benchmark is largely copy-pasted from the memcpy benchmark.
Prepare the memcpy file for merge with memset by extracting out a
generic function.
Signed-off-by: Rabin Vincent <rabin.vincent@axis.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Rabin Vincent <rabinv@axis.com>
Link: http://lkml.kernel.org/r/1417535441-3965-1-git-send-email-rabin.vincent@axis.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/bench/mem-memcpy.c | 182 ++++++++++++++++++++++++------------------
1 file changed, 104 insertions(+), 78 deletions(-)
diff --git a/tools/perf/bench/mem-memcpy.c b/tools/perf/bench/mem-memcpy.c
index 2465141..4a4493a 100644
--- a/tools/perf/bench/mem-memcpy.c
+++ b/tools/perf/bench/mem-memcpy.c
@@ -52,16 +52,18 @@ typedef void *(*memcpy_t)(void *, const void *, size_t);
struct routine {
const char *name;
const char *desc;
- memcpy_t fn;
+ union {
+ memcpy_t memcpy;
+ } fn;
};
-struct routine routines[] = {
- { "default",
- "Default memcpy() provided by glibc",
- memcpy },
+struct routine memcpy_routines[] = {
+ { .name = "default",
+ .desc = "Default memcpy() provided by glibc",
+ .fn.memcpy = memcpy },
#ifdef HAVE_ARCH_X86_64_SUPPORT
-#define MEMCPY_FN(fn, name, desc) { name, desc, fn },
+#define MEMCPY_FN(_fn, _name, _desc) {.name = _name, .desc = _desc, .fn.memcpy = _fn},
#include "mem-memcpy-x86-64-asm-def.h"
#undef MEMCPY_FN
@@ -69,7 +71,7 @@ struct routine routines[] = {
{ NULL,
NULL,
- NULL }
+ {NULL} }
};
static const char * const bench_mem_memcpy_usage[] = {
@@ -110,63 +112,6 @@ static double timeval2double(struct timeval *ts)
(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 { \
@@ -180,8 +125,16 @@ static double do_memcpy_gettimeofday(memcpy_t fn, size_t len, bool prefault)
printf(" %14lf GB/Sec", x / K / K / K); \
} while (0)
-int bench_mem_memcpy(int argc, const char **argv,
- const char *prefix __maybe_unused)
+struct bench_mem_info {
+ const struct routine *routines;
+ u64 (*do_cycle)(const struct routine *r, size_t len, bool prefault);
+ double (*do_gettimeofday)(const struct routine *r, size_t len, bool prefault);
+ const char *const *usage;
+};
+
+static int bench_mem_common(int argc, const char **argv,
+ const char *prefix __maybe_unused,
+ struct bench_mem_info *info)
{
int i;
size_t len;
@@ -189,7 +142,7 @@ int bench_mem_memcpy(int argc, const char **argv,
u64 result_cycle[2];
argc = parse_options(argc, argv, options,
- bench_mem_memcpy_usage, 0);
+ info->usage, 0);
if (no_prefault && only_prefault) {
fprintf(stderr, "Invalid options: -o and -n are mutually exclusive\n");
@@ -213,16 +166,16 @@ int bench_mem_memcpy(int argc, const char **argv,
if (only_prefault && no_prefault)
only_prefault = no_prefault = false;
- for (i = 0; routines[i].name; i++) {
- if (!strcmp(routines[i].name, routine))
+ for (i = 0; info->routines[i].name; i++) {
+ if (!strcmp(info->routines[i].name, routine))
break;
}
- if (!routines[i].name) {
+ if (!info->routines[i].name) {
printf("Unknown routine:%s\n", routine);
printf("Available routines...\n");
- for (i = 0; routines[i].name; i++) {
+ for (i = 0; info->routines[i].name; i++) {
printf("\t%s ... %s\n",
- routines[i].name, routines[i].desc);
+ info->routines[i].name, info->routines[i].desc);
}
return 1;
}
@@ -234,25 +187,25 @@ int bench_mem_memcpy(int argc, const char **argv,
/* show both of results */
if (use_cycle) {
result_cycle[0] =
- do_memcpy_cycle(routines[i].fn, len, false);
+ info->do_cycle(&info->routines[i], len, false);
result_cycle[1] =
- do_memcpy_cycle(routines[i].fn, len, true);
+ info->do_cycle(&info->routines[i], len, true);
} else {
result_bps[0] =
- do_memcpy_gettimeofday(routines[i].fn,
+ info->do_gettimeofday(&info->routines[i],
len, false);
result_bps[1] =
- do_memcpy_gettimeofday(routines[i].fn,
+ info->do_gettimeofday(&info->routines[i],
len, true);
}
} else {
if (use_cycle) {
result_cycle[pf] =
- do_memcpy_cycle(routines[i].fn,
+ info->do_cycle(&info->routines[i],
len, only_prefault);
} else {
result_bps[pf] =
- do_memcpy_gettimeofday(routines[i].fn,
+ info->do_gettimeofday(&info->routines[i],
len, only_prefault);
}
}
@@ -310,3 +263,76 @@ int bench_mem_memcpy(int argc, const char **argv,
return 0;
}
+
+static void memcpy_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(const struct routine *r, size_t len, bool prefault)
+{
+ u64 cycle_start = 0ULL, cycle_end = 0ULL;
+ void *src = NULL, *dst = NULL;
+ memcpy_t fn = r->fn.memcpy;
+ int i;
+
+ memcpy_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(const struct routine *r, size_t len,
+ bool prefault)
+{
+ struct timeval tv_start, tv_end, tv_diff;
+ memcpy_t fn = r->fn.memcpy;
+ void *src = NULL, *dst = NULL;
+ int i;
+
+ memcpy_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));
+}
+
+int bench_mem_memcpy(int argc, const char **argv,
+ const char *prefix __maybe_unused)
+{
+ struct bench_mem_info info = {
+ .routines = memcpy_routines,
+ .do_cycle = do_memcpy_cycle,
+ .do_gettimeofday = do_memcpy_gettimeofday,
+ .usage = bench_mem_memcpy_usage,
+ };
+
+ return bench_mem_common(argc, argv, prefix, &info);
+}
^ permalink raw reply related [flat|nested] 7+ messages in thread
* [tip:perf/urgent] perf bench: Merge memset into memcpy
2014-12-02 15:50 ` [PATCH 2/3] perf: bench: merge memset into memcpy Rabin Vincent
@ 2014-12-12 8:16 ` tip-bot for Rabin Vincent
0 siblings, 0 replies; 7+ messages in thread
From: tip-bot for Rabin Vincent @ 2014-12-12 8:16 UTC (permalink / raw)
To: linux-tip-commits
Cc: mingo, mingo, acme, linux-kernel, tglx, a.p.zijlstra, hpa, paulus,
rabin.vincent, rabinv
Commit-ID: 5bce1a5772cb52aad7e0466484ba07cfbfec2478
Gitweb: http://git.kernel.org/tip/5bce1a5772cb52aad7e0466484ba07cfbfec2478
Author: Rabin Vincent <rabin.vincent@axis.com>
AuthorDate: Tue, 2 Dec 2014 16:50:40 +0100
Committer: Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Tue, 9 Dec 2014 09:14:05 -0300
perf bench: Merge memset into memcpy
The memset benchmark is largely copy-pasted from the memcpy benchmark.
Merge the two now that memcpy is made more generic.
Signed-off-by: Rabin Vincent <rabin.vincent@axis.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Rabin Vincent <rabinv@axis.com>
Link: http://lkml.kernel.org/r/1417535441-3965-2-git-send-email-rabin.vincent@axis.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/Makefile.perf | 1 -
tools/perf/bench/mem-memcpy.c | 90 +++++++++++++
tools/perf/bench/mem-memset.c | 304 ------------------------------------------
3 files changed, 90 insertions(+), 305 deletions(-)
diff --git a/tools/perf/Makefile.perf b/tools/perf/Makefile.perf
index 478efa9..763e68f 100644
--- a/tools/perf/Makefile.perf
+++ b/tools/perf/Makefile.perf
@@ -458,7 +458,6 @@ BUILTIN_OBJS += $(OUTPUT)bench/mem-memcpy-x86-64-asm.o
BUILTIN_OBJS += $(OUTPUT)bench/mem-memset-x86-64-asm.o
endif
BUILTIN_OBJS += $(OUTPUT)bench/mem-memcpy.o
-BUILTIN_OBJS += $(OUTPUT)bench/mem-memset.o
BUILTIN_OBJS += $(OUTPUT)bench/futex-hash.o
BUILTIN_OBJS += $(OUTPUT)bench/futex-wake.o
BUILTIN_OBJS += $(OUTPUT)bench/futex-requeue.o
diff --git a/tools/perf/bench/mem-memcpy.c b/tools/perf/bench/mem-memcpy.c
index 4a4493a..e18be70 100644
--- a/tools/perf/bench/mem-memcpy.c
+++ b/tools/perf/bench/mem-memcpy.c
@@ -13,6 +13,7 @@
#include "../util/cloexec.h"
#include "bench.h"
#include "mem-memcpy-arch.h"
+#include "mem-memset-arch.h"
#include <stdio.h>
#include <stdlib.h>
@@ -48,12 +49,14 @@ static const struct option options[] = {
};
typedef void *(*memcpy_t)(void *, const void *, size_t);
+typedef void *(*memset_t)(void *, int, size_t);
struct routine {
const char *name;
const char *desc;
union {
memcpy_t memcpy;
+ memset_t memset;
} fn;
};
@@ -336,3 +339,90 @@ int bench_mem_memcpy(int argc, const char **argv,
return bench_mem_common(argc, argv, prefix, &info);
}
+
+static void memset_alloc_mem(void **dst, size_t length)
+{
+ *dst = zalloc(length);
+ if (!*dst)
+ die("memory allocation failed - maybe length is too large?\n");
+}
+
+static u64 do_memset_cycle(const struct routine *r, size_t len, bool prefault)
+{
+ u64 cycle_start = 0ULL, cycle_end = 0ULL;
+ memset_t fn = r->fn.memset;
+ void *dst = NULL;
+ int i;
+
+ memset_alloc_mem(&dst, len);
+
+ if (prefault)
+ fn(dst, -1, len);
+
+ cycle_start = get_cycle();
+ for (i = 0; i < iterations; ++i)
+ fn(dst, i, len);
+ cycle_end = get_cycle();
+
+ free(dst);
+ return cycle_end - cycle_start;
+}
+
+static double do_memset_gettimeofday(const struct routine *r, size_t len,
+ bool prefault)
+{
+ struct timeval tv_start, tv_end, tv_diff;
+ memset_t fn = r->fn.memset;
+ void *dst = NULL;
+ int i;
+
+ memset_alloc_mem(&dst, len);
+
+ if (prefault)
+ fn(dst, -1, len);
+
+ BUG_ON(gettimeofday(&tv_start, NULL));
+ for (i = 0; i < iterations; ++i)
+ fn(dst, i, len);
+ BUG_ON(gettimeofday(&tv_end, NULL));
+
+ timersub(&tv_end, &tv_start, &tv_diff);
+
+ free(dst);
+ return (double)((double)len / timeval2double(&tv_diff));
+}
+
+static const char * const bench_mem_memset_usage[] = {
+ "perf bench mem memset <options>",
+ NULL
+};
+
+static const struct routine memset_routines[] = {
+ { .name ="default",
+ .desc = "Default memset() provided by glibc",
+ .fn.memset = memset },
+#ifdef HAVE_ARCH_X86_64_SUPPORT
+
+#define MEMSET_FN(_fn, _name, _desc) { .name = _name, .desc = _desc, .fn.memset = _fn },
+#include "mem-memset-x86-64-asm-def.h"
+#undef MEMSET_FN
+
+#endif
+
+ { .name = NULL,
+ .desc = NULL,
+ .fn.memset = NULL }
+};
+
+int bench_mem_memset(int argc, const char **argv,
+ const char *prefix __maybe_unused)
+{
+ struct bench_mem_info info = {
+ .routines = memset_routines,
+ .do_cycle = do_memset_cycle,
+ .do_gettimeofday = do_memset_gettimeofday,
+ .usage = bench_mem_memset_usage,
+ };
+
+ return bench_mem_common(argc, argv, prefix, &info);
+}
diff --git a/tools/perf/bench/mem-memset.c b/tools/perf/bench/mem-memset.c
deleted file mode 100644
index 75fc3e6..0000000
--- a/tools/perf/bench/mem-memset.c
+++ /dev/null
@@ -1,304 +0,0 @@
-/*
- * mem-memset.c
- *
- * memset: Simple memory set in various ways
- *
- * Trivial clone of mem-memcpy.c.
- */
-
-#include "../perf.h"
-#include "../util/util.h"
-#include "../util/parse-options.h"
-#include "../util/header.h"
-#include "../util/cloexec.h"
-#include "bench.h"
-#include "mem-memset-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 set. "
- "Available units: B, KB, MB, GB and TB (upper and lower)"),
- OPT_STRING('r', "routine", &routine, "default",
- "Specify routine to set"),
- OPT_INTEGER('i', "iterations", &iterations,
- "repeat memset() 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 memset()"),
- OPT_BOOLEAN('n', "no-prefault", &no_prefault,
- "Show only the result without page faults before memset()"),
- OPT_END()
-};
-
-typedef void *(*memset_t)(void *, int, size_t);
-
-struct routine {
- const char *name;
- const char *desc;
- memset_t fn;
-};
-
-static const struct routine routines[] = {
- { "default",
- "Default memset() provided by glibc",
- memset },
-#ifdef HAVE_ARCH_X86_64_SUPPORT
-
-#define MEMSET_FN(fn, name, desc) { name, desc, fn },
-#include "mem-memset-x86-64-asm-def.h"
-#undef MEMSET_FN
-
-#endif
-
- { NULL,
- NULL,
- NULL }
-};
-
-static const char * const bench_mem_memset_usage[] = {
- "perf bench mem memset <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,
- perf_event_open_cloexec_flag());
-
- 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, size_t length)
-{
- *dst = zalloc(length);
- if (!*dst)
- die("memory allocation failed - maybe length is too large?\n");
-}
-
-static u64 do_memset_cycle(memset_t fn, size_t len, bool prefault)
-{
- u64 cycle_start = 0ULL, cycle_end = 0ULL;
- void *dst = NULL;
- int i;
-
- alloc_mem(&dst, len);
-
- if (prefault)
- fn(dst, -1, len);
-
- cycle_start = get_cycle();
- for (i = 0; i < iterations; ++i)
- fn(dst, i, len);
- cycle_end = get_cycle();
-
- free(dst);
- return cycle_end - cycle_start;
-}
-
-static double do_memset_gettimeofday(memset_t fn, size_t len, bool prefault)
-{
- struct timeval tv_start, tv_end, tv_diff;
- void *dst = NULL;
- int i;
-
- alloc_mem(&dst, len);
-
- if (prefault)
- fn(dst, -1, len);
-
- BUG_ON(gettimeofday(&tv_start, NULL));
- for (i = 0; i < iterations; ++i)
- fn(dst, i, len);
- BUG_ON(gettimeofday(&tv_end, NULL));
-
- timersub(&tv_end, &tv_start, &tv_diff);
-
- 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_memset(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_memset_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_memset_cycle(routines[i].fn, len, false);
- result_cycle[1] =
- do_memset_cycle(routines[i].fn, len, true);
- } else {
- result_bps[0] =
- do_memset_gettimeofday(routines[i].fn,
- len, false);
- result_bps[1] =
- do_memset_gettimeofday(routines[i].fn,
- len, true);
- }
- } else {
- if (use_cycle) {
- result_cycle[pf] =
- do_memset_cycle(routines[i].fn,
- len, only_prefault);
- } else {
- result_bps[pf] =
- do_memset_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;
-}
^ permalink raw reply related [flat|nested] 7+ messages in thread
* [tip:perf/urgent] perf bench: Fix memcpy/memset output
2014-12-02 15:50 ` [PATCH 3/3] perf: bench: fix memcpy/memset output Rabin Vincent
2014-12-08 10:25 ` Ingo Molnar
@ 2014-12-12 8:16 ` tip-bot for Rabin Vincent
1 sibling, 0 replies; 7+ messages in thread
From: tip-bot for Rabin Vincent @ 2014-12-12 8:16 UTC (permalink / raw)
To: linux-tip-commits
Cc: linux-kernel, mingo, tglx, acme, paulus, a.p.zijlstra, mingo,
rabin, rabin.vincent, hpa, rabinv
Commit-ID: 1182f883113483cefbc3be0178a2df2dc9ae8b77
Gitweb: http://git.kernel.org/tip/1182f883113483cefbc3be0178a2df2dc9ae8b77
Author: Rabin Vincent <rabin.vincent@axis.com>
AuthorDate: Tue, 2 Dec 2014 16:50:41 +0100
Committer: Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Tue, 9 Dec 2014 09:14:08 -0300
perf bench: Fix memcpy/memset output
The memcpy and memset benchmarks return bogus results when iterations >
0 because the iterations value is not taken into account when
calculating the final result:
$ perf bench mem memset --only-prefault --length 1GB --iterations 1
# Running 'mem/memset' benchmark:
# Copying 1GB Bytes ...
20.798669 GB/Sec (with prefault)
$ perf bench mem memset --only-prefault --length 1GB --iterations 10
# Running 'mem/memset' benchmark:
# Copying 1GB Bytes ...
2.086576 GB/Sec (with prefault)
$ perf bench mem memset --only-prefault --length 1GB --iterations 100
# Running 'mem/memset' benchmark:
# Copying 1GB Bytes ...
212.840917 MB/Sec (with prefault)
Fix this.
Signed-off-by: Rabin Vincent <rabin.vincent@axis.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Rabin Vincent <rabin@rab.in>
Cc: Rabin Vincent <rabinv@axis.com>
Link: http://lkml.kernel.org/r/1417535441-3965-3-git-send-email-rabin.vincent@axis.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/bench/mem-memcpy.c | 18 ++++++++++--------
1 file changed, 10 insertions(+), 8 deletions(-)
diff --git a/tools/perf/bench/mem-memcpy.c b/tools/perf/bench/mem-memcpy.c
index e18be70..6c14afe 100644
--- a/tools/perf/bench/mem-memcpy.c
+++ b/tools/perf/bench/mem-memcpy.c
@@ -141,6 +141,7 @@ static int bench_mem_common(int argc, const char **argv,
{
int i;
size_t len;
+ double totallen;
double result_bps[2];
u64 result_cycle[2];
@@ -156,6 +157,7 @@ static int bench_mem_common(int argc, const char **argv,
init_cycle();
len = (size_t)perf_atoll((char *)length_str);
+ totallen = (double)len * iterations;
result_cycle[0] = result_cycle[1] = 0ULL;
result_bps[0] = result_bps[1] = 0.0;
@@ -219,10 +221,10 @@ static int bench_mem_common(int argc, const char **argv,
if (use_cycle) {
printf(" %14lf Cycle/Byte\n",
(double)result_cycle[0]
- / (double)len);
+ / totallen);
printf(" %14lf Cycle/Byte (with prefault)\n",
(double)result_cycle[1]
- / (double)len);
+ / totallen);
} else {
print_bps(result_bps[0]);
printf("\n");
@@ -233,7 +235,7 @@ static int bench_mem_common(int argc, const char **argv,
if (use_cycle) {
printf(" %14lf Cycle/Byte",
(double)result_cycle[pf]
- / (double)len);
+ / totallen);
} else
print_bps(result_bps[pf]);
@@ -244,8 +246,8 @@ static int bench_mem_common(int argc, const char **argv,
if (!only_prefault && !no_prefault) {
if (use_cycle) {
printf("%lf %lf\n",
- (double)result_cycle[0] / (double)len,
- (double)result_cycle[1] / (double)len);
+ (double)result_cycle[0] / totallen,
+ (double)result_cycle[1] / totallen);
} else {
printf("%lf %lf\n",
result_bps[0], result_bps[1]);
@@ -253,7 +255,7 @@ static int bench_mem_common(int argc, const char **argv,
} else {
if (use_cycle) {
printf("%lf\n", (double)result_cycle[pf]
- / (double)len);
+ / totallen);
} else
printf("%lf\n", result_bps[pf]);
}
@@ -324,7 +326,7 @@ static double do_memcpy_gettimeofday(const struct routine *r, size_t len,
free(src);
free(dst);
- return (double)((double)len / timeval2double(&tv_diff));
+ return (double)(((double)len * iterations) / timeval2double(&tv_diff));
}
int bench_mem_memcpy(int argc, const char **argv,
@@ -389,7 +391,7 @@ static double do_memset_gettimeofday(const struct routine *r, size_t len,
timersub(&tv_end, &tv_start, &tv_diff);
free(dst);
- return (double)((double)len / timeval2double(&tv_diff));
+ return (double)(((double)len * iterations) / timeval2double(&tv_diff));
}
static const char * const bench_mem_memset_usage[] = {
^ permalink raw reply related [flat|nested] 7+ messages in thread
end of thread, other threads:[~2014-12-12 8:16 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2014-12-02 15:50 [PATCH 1/3] perf: bench: prepare memcpy for merge Rabin Vincent
2014-12-02 15:50 ` [PATCH 2/3] perf: bench: merge memset into memcpy Rabin Vincent
2014-12-12 8:16 ` [tip:perf/urgent] perf bench: Merge " tip-bot for Rabin Vincent
2014-12-02 15:50 ` [PATCH 3/3] perf: bench: fix memcpy/memset output Rabin Vincent
2014-12-08 10:25 ` Ingo Molnar
2014-12-12 8:16 ` [tip:perf/urgent] perf bench: Fix " tip-bot for Rabin Vincent
2014-12-12 8:15 ` [tip:perf/urgent] perf bench: Prepare memcpy for merge tip-bot for Rabin Vincent
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.