LinuxLists.cc - [PATCH 0/1] perf stat: Add support to print counts for fixed times

2018-01-17 12:37:44

Subject: [PATCH 0/1] perf stat: Add support to print counts for fixed times

2018-01-17 12:37:52

Subject: [PATCH 1/1] perf stat: Add support to print counts for fixed times

From: yuzhoujian <[email protected]>

Introduce a new option to print counts for fixed number of times
and update perf-stat documentation accordingly.

Show below is the output of the new option for perf stat.

$perf stat -I 1000 --times-print 2 -e cycles -a
# time counts unit events
1.002827089 93,884,870 cycles
2.004231506 56,573,446 cycles

We can just print the counts for several times with this newly introduced
option. The usage of it is a little like vmstat.

$ vmstat -n 1 2
procs -----------memory---------- ---swap-- -----io---- -system-- ------cpu-----
r b swpd free buff cache si so bi bo in cs us sy id wa st
0 0 0 78270544 547484 51732076 0 0 0 20 1 1 1 0 99 0 0
0 0 0 78270512 547484 51732080 0 0 0 16 477 1555 0 0 100 0 0

Signed-off-by: yuzhoujian <[email protected]>
---
tools/perf/Documentation/perf-stat.txt | 5 +++++
tools/perf/builtin-stat.c | 13 +++++++++++++
tools/perf/util/stat.h | 1 +
3 files changed, 19 insertions(+)

diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
index 823fce7674bb..8c4526c287c4 100644
--- a/tools/perf/Documentation/perf-stat.txt
+++ b/tools/perf/Documentation/perf-stat.txt
@@ -146,6 +146,11 @@ Print count deltas every N milliseconds (minimum: 10ms)
The overhead percentage could be high in some cases, for instance with small, sub 100ms intervals. Use with caution.
example: 'perf stat -I 1000 -e cycles -a sleep 5'

+--times-print::
+Print count deltas for fixed number of times.
+This option should be used together with "-I" option.
+ example: 'perf stat -I 1000 --times-print 2 -e cycles -a'
+
--metric-only::
Only print computed metrics. Print them in a single line.
Don't show any raw values. Not supported with --per-thread.
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 98bf9d32f222..47459109110f 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -168,6 +168,7 @@ static struct timespec ref_time;
static struct cpu_map *aggr_map;
static aggr_get_id_t aggr_get_id;
static bool append_file;
+static bool times_print;
static const char *output_name;
static int output_fd;
static int print_free_counters_hint;
@@ -700,6 +701,10 @@ static int __run_perf_stat(int argc, const char **argv)
while (!waitpid(child_pid, &status, WNOHANG)) {
nanosleep(&ts, NULL);
process_interval();
+ if (times_print == true) {
+ if (--stat_config.times == 0)
+ break;
+ }
}
}
waitpid(child_pid, &status, 0);
@@ -718,6 +723,10 @@ static int __run_perf_stat(int argc, const char **argv)
nanosleep(&ts, NULL);
if (interval)
process_interval();
+ if (times_print == true) {
+ if (--stat_config.times == 0)
+ break;
+ }
}
}

@@ -1891,6 +1900,8 @@ static const struct option stat_options[] = {
"command to run after to the measured command"),
OPT_UINTEGER('I', "interval-print", &stat_config.interval,
"print counts at regular interval in ms (>= 10)"),
+ OPT_INTEGER(0, "times-print", &stat_config.times,
+ "print counts for fixed number of times"),
OPT_SET_UINT(0, "per-socket", &stat_config.aggr_mode,
"aggregate counts per processor socket", AGGR_SOCKET),
OPT_SET_UINT(0, "per-core", &stat_config.aggr_mode,
@@ -2870,6 +2881,8 @@ int cmd_stat(int argc, const char **argv)
"The overhead percentage could be high in some cases. "
"Please proceed with caution.\n");
}
+ if (stat_config.times)
+ times_print = true;

if (perf_evlist__alloc_stats(evsel_list, interval))
goto out;
diff --git a/tools/perf/util/stat.h b/tools/perf/util/stat.h
index dbc6f7134f61..b2f8a347d358 100644
--- a/tools/perf/util/stat.h
+++ b/tools/perf/util/stat.h
@@ -90,6 +90,7 @@ struct perf_stat_config {
bool scale;
FILE *output;
unsigned int interval;
+ int times;
struct runtime_stat *stats;
int stats_num;
};
--
2.14.1

2018-01-23 14:25:42

by Jiri Olsa

[permalink] [raw]

Subject: Re: [PATCH 1/1] perf stat: Add support to print counts for fixed times

On Wed, Jan 17, 2018 at 01:37:25PM +0100, ufo19890607 wrote:
> From: yuzhoujian <[email protected]>
>
> Introduce a new option to print counts for fixed number of times
> and update perf-stat documentation accordingly.
>
> Show below is the output of the new option for perf stat.
>
> $perf stat -I 1000 --times-print 2 -e cycles -a
> # time counts unit events
> 1.002827089 93,884,870 cycles
> 2.004231506 56,573,446 cycles

you can do that with following command:

[jolsa@krava perf]$ sudo ./perf stat -I 1000 -e cycles -a sleep 2
# time counts unit events
1.000160132 437,471,026 cycles
2.000367121 729,415,520 cycles
2.000857390 1,407,820 cycles

jirka

2018-01-24 07:40:11

by Jiri Olsa

[permalink] [raw]

Subject: Re: [PATCH 1/1] perf stat: Add support to print counts for fixed times

On Wed, Jan 24, 2018 at 10:03:08AM +0800, 禹舟键 wrote:
> Hi Jirka
> How can you control the output for perf stat, if the "--interval-print" is
> not 1000, eg. 2000, 3000, 5000.
>
> root@node10:/home$ ./perf stat -e cycles -a -I 2000 sleep 3
> # time counts unit events
> 2.001361101 695,569,361 cycles
> 3.003249250 57,904,006 cycles
> root@node10:/home$ ./perf stat -e cycles -a -I 2000 --times-print 3
> # time counts unit events
> 2.002784924 128,438,406
> cycles
> 4.004290885 93,128,097
> cycles
> 6.005726281 111,303,359 cycles

[jolsa@krava perf]$ sudo ./perf stat -e cycles -a -I 2000 --times-print 3
# time counts unit events
2.000161257 1,197,889,478 cycles
4.000424645 2,905,635,530 cycles
6.000589943 1,658,546,001 cycles
[jolsa@krava perf]$ sudo ./perf stat -e cycles -a -I 2000 sleep 6
# time counts unit events
2.000248489 1,224,257,116 cycles
4.000489261 3,171,435,328 cycles
6.000766440 4,092,029,501 cycles
6.916537222 563,745,072 cycles

jirka