perf stat: Improve runtime stat for interval mode
PATCH 1/1 perf stat: Improve runtime stat for interval mode
For interval mode, the metric is printed after # if it exists. But it's not calculated by the counts generated in this interval. See following examples,
root@kbl-ppc:~# perf stat -M CPI -I1000 --interval-count 2
# time counts unit events
1.000422803 764,809 inst_retired.any # 2.9 CPI
1.000422803 2,234,932 cycles
2.001464585 1,960,061 inst_retired.any # 1.6 CPI
2.001464585 4,022,591 cyclesThe second CPI should not be 1.6 (4,022,591/1,960,061 is 2.1)
root@kbl-ppc:~# perf stat -e cycles,instructions -I1000 --interval-count 2
# time counts unit events
1.000429493 2,869,311 cycles
1.000429493 816,875 instructions # 0.28 insn per cycle
2.001516426 9,260,973 cycles
2.001516426 5,250,634 instructions # 0.87 insn per cycleThe second 'insn per cycle' should not be 0.87 (5,250,634/9,260,973 is 0.57).
The current code uses a global variable rt_stat for tracking and updating the std dev of runtime stat. Unlike the counts, rt_stat is not reset for interval. While the counts are reset for interval.
perf_stat_process_counter()
{
if (config->interval)
init_stats(ps->res_stats);
}So for interval, the rt_stat should be reset either.
This patch resets rt_stat before read_counters, so the runtime stat is only calculated by the counts generated in this interval.
With this patch,
root@kbl-ppc:~# perf stat -M CPI -I1000 --interval-count 2
# time counts unit events
1.000420924 2,408,818 inst_retired.any # 2.1 CPI
1.000420924 5,010,111 cycles
2.001448579 2,798,407 inst_retired.any # 1.6 CPI
2.001448579 4,599,861 cycles root@kbl-ppc:~# perf stat -e cycles,instructions -I1000 --interval-count 2
# time counts unit events
1.000428555 2,769,714 cycles
1.000428555 774,462 instructions # 0.28 insn per cycle
2.001471562 3,595,904 cycles
2.001471562 1,243,703 instructions # 0.35 insn per cycleNow the second 'insn per cycle' and CPI are calculated by the counts generated in this interval.
--- tools/perf/Documentation/perf-stat.txt | 2 ++ tools/perf/builtin-stat.c | 1 + tools/perf/util/stat-shadow.c | 5 +++++ tools/perf/util/stat.h | 1 + 4 files changed, 9 insertions(+)
tools/perf/Documentation/perf-stat.txt +2 -0
@@ -176,6 +176,8 @@ Print count deltas every N milliseconds (minimum: 1ms) 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' +If the metric exists, it is calculated by the counts generated in this interval and the metric is printed after #. + --interval-count times:: Print count deltas for fixed number of times. This option should be used together with "-I" option.
tools/perf/builtin-stat.c +1 -0
@@ -359,6 +359,7 @@ static void process_interval(void)clock_gettime (CLOCK_MONOTONIC , &ts);diff_timespec (&rs, &ts, &ref_time); +perf_stat__reset_rt_stat ();read_counters (&rs);if (STAT_RECORD ) {
tools/perf/util/stat-shadow.c +5 -0
@@ -186,6 +186,11 @@ void perf_stat__reset_shadow_stats(void)memset (&walltime_nsecs_stats,0 ,sizeof (walltime_nsecs_stats)); } +void perf_stat__reset_rt_stat (void ) +{ +reset_stat (&rt_stat); +} +void perf_stat__reset_shadow_per_stat (struct runtime_stat *st) {reset_stat (st);
tools/perf/util/stat.h +1 -0
@@ -176,6 +176,7 @@ void runtime_stat__init(struct runtime_stat *st);void runtime_stat__exit (struct runtime_stat *st);void perf_stat__init_shadow_stats (void );void perf_stat__reset_shadow_stats (void ); +void perf_stat__reset_rt_stat (void );void perf_stat__reset_shadow_per_stat (struct runtime_stat *st);void perf_stat__update_shadow_stats (struct evsel *counter,u64 count,int cpu,struct runtime_stat *st);
Signature · 1 line
2.17.1
Quoted 46 lines · Jin Yao
For interval mode, the metric is printed after # if it exists. But it's not calculated by the counts generated in this interval. See following examples,
root@kbl-ppc:~# perf stat -M CPI -I1000 --interval-count 2 # time counts unit events 1.000422803 764,809 inst_retired.any # 2.9 CPI 1.000422803 2,234,932 cycles 2.001464585 1,960,061 inst_retired.any # 1.6 CPI 2.001464585 4,022,591 cyclesThe second CPI should not be 1.6 (4,022,591/1,960,061 is 2.1)
root@kbl-ppc:~# perf stat -e cycles,instructions -I1000 --interval-count 2 # time counts unit events 1.000429493 2,869,311 cycles 1.000429493 816,875 instructions # 0.28 insn per cycle 2.001516426 9,260,973 cycles 2.001516426 5,250,634 instructions # 0.87 insn per cycleThe second 'insn per cycle' should not be 0.87 (5,250,634/9,260,973 is 0.57).
The current code uses a global variable
rt_statfor tracking and updating the std dev of runtime stat. Unlike the counts,rt_statis not reset for interval. While the counts are reset for interval.
perf_stat_process_counter(){ if (config->interval) init_stats(ps->res_stats); }So for interval, the
rt_statshould be reset either.This patch resets
rt_statbeforeread_counters, so the runtime stat is only calculated by the counts generated in this interval.With this patch,
root@kbl-ppc:~# perf stat -M CPI -I1000 --interval-count 2 # time counts unit events 1.000420924 2,408,818 inst_retired.any # 2.1 CPI 1.000420924 5,010,111 cycles 2.001448579 2,798,407 inst_retired.any # 1.6 CPI 2.001448579 4,599,861 cycles
Quoted 7 lines
root@kbl-ppc:~# perf stat -e cycles,instructions -I1000 --interval-count 2
# time counts unit events
1.000428555 2,769,714 cycles
1.000428555 774,462 instructions # 0.28 insn per cycle
2.001471562 3,595,904 cycles
2.001471562 1,243,703 instructions # 0.35 insn per cycleHi Jin,
Thanks for the patch. Was working on similar issue raised by Joakim Zhang.
Here's link to my findings: https://lkml.org/lkml/2020/3/24/158
This patch solves the issue.
Quoted 64 lines
Now the second 'insn per cycle' and CPI are calculated by the counts generated in this interval.>
Signed-off-by Jin Yao <[email protected]>--- tools/perf/Documentation/perf-stat.txt | 2 ++ tools/perf/builtin-stat.c | 1 + tools/perf/util/stat-shadow.c | 5 +++++ tools/perf/util/stat.h | 1 + 4 files changed, 9 insertions(+)tools/perf/Documentation/perf-stat.txt +2
-0@@ -176,6 +176,8 @@ Print count deltas every N milliseconds (minimum: 1ms) 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' +If the metric exists, it is calculated by the counts generated in this interval and the metric is printed after #. + --interval-count times:: Print count deltas for fixed number of times. This option should be used together with "-I" option.tools/perf/builtin-stat.c +1
-0@@ -359,6 +359,7 @@ static void process_interval(void)clock_gettime (CLOCK_MONOTONIC , &ts);diff_timespec (&rs, &ts, &ref_time); +perf_stat__reset_rt_stat ();read_counters (&rs);if (STAT_RECORD ) {tools/perf/util/stat-shadow.c +5
-0@@ -186,6 +186,11 @@ void perf_stat__reset_shadow_stats(void)memset (&walltime_nsecs_stats,0 ,sizeof (walltime_nsecs_stats)); } +void perf_stat__reset_rt_stat (void ) +{ +reset_stat (&rt_stat); +} +void perf_stat__reset_shadow_per_stat (struct runtime_stat *st) {reset_stat (st);tools/perf/util/stat.h +1
-0@@ -176,6 +176,7 @@ void runtime_stat__init(struct runtime_stat *st);void runtime_stat__exit (struct runtime_stat *st);void perf_stat__init_shadow_stats (void );void perf_stat__reset_shadow_stats (void ); +void perf_stat__reset_rt_stat (void );void perf_stat__reset_shadow_per_stat (struct runtime_stat *st);void perf_stat__update_shadow_stats (struct evsel *counter,u64 count,int cpu,struct runtime_stat *st);
Quoted 46 lines · Jin Yao
For interval mode, the metric is printed after # if it exists. But it's not calculated by the counts generated in this interval. See following examples,
root@kbl-ppc:~# perf stat -M CPI -I1000 --interval-count 2 # time counts unit events 1.000422803 764,809 inst_retired.any # 2.9 CPI 1.000422803 2,234,932 cycles 2.001464585 1,960,061 inst_retired.any # 1.6 CPI 2.001464585 4,022,591 cyclesThe second CPI should not be 1.6 (4,022,591/1,960,061 is 2.1)
root@kbl-ppc:~# perf stat -e cycles,instructions -I1000 --interval-count 2 # time counts unit events 1.000429493 2,869,311 cycles 1.000429493 816,875 instructions # 0.28 insn per cycle 2.001516426 9,260,973 cycles 2.001516426 5,250,634 instructions # 0.87 insn per cycleThe second 'insn per cycle' should not be 0.87 (5,250,634/9,260,973 is 0.57).
The current code uses a global variable
rt_statfor tracking and updating the std dev of runtime stat. Unlike the counts,rt_statis not reset for interval. While the counts are reset for interval.
perf_stat_process_counter(){ if (config->interval) init_stats(ps->res_stats); }So for interval, the
rt_statshould be reset either.This patch resets
rt_statbeforeread_counters, so the runtime stat is only calculated by the counts generated in this interval.With this patch,
root@kbl-ppc:~# perf stat -M CPI -I1000 --interval-count 2 # time counts unit events 1.000420924 2,408,818 inst_retired.any # 2.1 CPI 1.000420924 5,010,111 cycles 2.001448579 2,798,407 inst_retired.any # 1.6 CPI 2.001448579 4,599,861 cycles
Quoted 7 lines
root@kbl-ppc:~# perf stat -e cycles,instructions -I1000 --interval-count 2
# time counts unit events
1.000428555 2,769,714 cycles
1.000428555 774,462 instructions # 0.28 insn per cycle
2.001471562 3,595,904 cycles
2.001471562 1,243,703 instructions # 0.35 insn per cycleHi Jin,
Thanks for the patch. Was working on similar issue raised by Joakim Zhang.
Here's link to my findings: https://lkml.org/lkml/2020/3/24/158
This patch solves the issue.
Quoted 65 lines
Now the second 'insn per cycle' and CPI are calculated by the counts generated in this interval.
Signed-off-by Jin Yao <[email protected]>--- tools/perf/Documentation/perf-stat.txt | 2 ++ tools/perf/builtin-stat.c | 1 + tools/perf/util/stat-shadow.c | 5 +++++ tools/perf/util/stat.h | 1 + 4 files changed, 9 insertions(+)tools/perf/Documentation/perf-stat.txt +2
-0@@ -176,6 +176,8 @@ Print count deltas every N milliseconds (minimum: 1ms) 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' +If the metric exists, it is calculated by the counts generated in this interval and the metric is printed after #. + --interval-count times:: Print count deltas for fixed number of times. This option should be used together with "-I" option.tools/perf/builtin-stat.c +1
-0@@ -359,6 +359,7 @@ static void process_interval(void)clock_gettime (CLOCK_MONOTONIC , &ts);diff_timespec (&rs, &ts, &ref_time); +perf_stat__reset_rt_stat ();read_counters (&rs);if (STAT_RECORD ) {tools/perf/util/stat-shadow.c +5
-0@@ -186,6 +186,11 @@ void perf_stat__reset_shadow_stats(void)memset (&walltime_nsecs_stats,0 ,sizeof (walltime_nsecs_stats)); } +void perf_stat__reset_rt_stat (void ) +{ +reset_stat (&rt_stat); +} +void perf_stat__reset_shadow_per_stat (struct runtime_stat *st) {reset_stat (st);tools/perf/util/stat.h +1
-0@@ -176,6 +176,7 @@ void runtime_stat__init(struct runtime_stat *st);void runtime_stat__exit (struct runtime_stat *st);void perf_stat__init_shadow_stats (void );void perf_stat__reset_shadow_stats (void ); +void perf_stat__reset_rt_stat (void );void perf_stat__reset_shadow_per_stat (struct runtime_stat *st);void perf_stat__update_shadow_stats (struct evsel *counter,u64 count,int cpu,struct runtime_stat *st);
Quoted 57 lines · kajoljain
Quoted 46 lines · Jin Yao
For interval mode, the metric is printed after # if it exists. But it's not calculated by the counts generated in this interval. See following examples,
root@kbl-ppc:~# perf stat -M CPI -I1000 --interval-count 2 # time counts unit events 1.000422803 764,809 inst_retired.any # 2.9 CPI 1.000422803 2,234,932 cycles 2.001464585 1,960,061 inst_retired.any # 1.6 CPI 2.001464585 4,022,591 cyclesThe second CPI should not be 1.6 (4,022,591/1,960,061 is 2.1)
root@kbl-ppc:~# perf stat -e cycles,instructions -I1000 --interval-count 2 # time counts unit events 1.000429493 2,869,311 cycles 1.000429493 816,875 instructions # 0.28 insn per cycle 2.001516426 9,260,973 cycles 2.001516426 5,250,634 instructions # 0.87 insn per cycleThe second 'insn per cycle' should not be 0.87 (5,250,634/9,260,973 is 0.57).
The current code uses a global variable
rt_statfor tracking and updating the std dev of runtime stat. Unlike the counts,rt_statis not reset for interval. While the counts are reset for interval.
perf_stat_process_counter(){ if (config->interval) init_stats(ps->res_stats); }So for interval, the
rt_statshould be reset either.This patch resets
rt_statbeforeread_counters, so the runtime stat is only calculated by the counts generated in this interval.With this patch,
root@kbl-ppc:~# perf stat -M CPI -I1000 --interval-count 2 # time counts unit events 1.000420924 2,408,818 inst_retired.any # 2.1 CPI 1.000420924 5,010,111 cycles 2.001448579 2,798,407 inst_retired.any # 1.6 CPI 2.001448579 4,599,861 cyclesQuoted 7 lines
root@kbl-ppc:~# perf stat -e cycles,instructions -I1000 --interval-count 2 # time counts unit events 1.000428555 2,769,714 cycles 1.000428555 774,462 instructions # 0.28 insn per cycle 2.001471562 3,595,904 cycles 2.001471562 1,243,703 instructions # 0.35 insn per cycle
Quoted 8 lines
Hi Jin,
Thanks for the patch. Was working on similar issue raised by Joakim Zhang.
Here's link to my findings: https://lkml.org/lkml/2020/3/24/158This patch solves the issue.
Thanks for testing this patch. :)
Thanks
Jin Yao
Quoted 65 lines
Quoted 65 lines
Now the second 'insn per cycle' and CPI are calculated by the counts generated in this interval.
Signed-off-by Jin Yao <[email protected]>--- tools/perf/Documentation/perf-stat.txt | 2 ++ tools/perf/builtin-stat.c | 1 + tools/perf/util/stat-shadow.c | 5 +++++ tools/perf/util/stat.h | 1 + 4 files changed, 9 insertions(+)tools/perf/Documentation/perf-stat.txt +2
-0@@ -176,6 +176,8 @@ Print count deltas every N milliseconds (minimum: 1ms) 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' +If the metric exists, it is calculated by the counts generated in this interval and the metric is printed after #. + --interval-count times:: Print count deltas for fixed number of times. This option should be used together with "-I" option.tools/perf/builtin-stat.c +1
-0@@ -359,6 +359,7 @@ static void process_interval(void)clock_gettime (CLOCK_MONOTONIC , &ts);diff_timespec (&rs, &ts, &ref_time); +perf_stat__reset_rt_stat ();read_counters (&rs);if (STAT_RECORD ) {tools/perf/util/stat-shadow.c +5
-0@@ -186,6 +186,11 @@ void perf_stat__reset_shadow_stats(void)memset (&walltime_nsecs_stats,0 ,sizeof (walltime_nsecs_stats)); } +void perf_stat__reset_rt_stat (void ) +{ +reset_stat (&rt_stat); +} +void perf_stat__reset_shadow_per_stat (struct runtime_stat *st) {reset_stat (st);tools/perf/util/stat.h +1
-0@@ -176,6 +176,7 @@ void runtime_stat__init(struct runtime_stat *st);void runtime_stat__exit (struct runtime_stat *st);void perf_stat__init_shadow_stats (void );void perf_stat__reset_shadow_stats (void ); +void perf_stat__reset_rt_stat (void );void perf_stat__reset_shadow_per_stat (struct runtime_stat *st);void perf_stat__update_shadow_stats (struct evsel *counter,u64 count,int cpu,struct runtime_stat *st);
On Fri, Apr 17, 2020 at 08:51:54AM +0800, Jin Yao wrote:
SNIP
tools/perf/builtin-stat.c static void process_interval(void) · line 362
Quoted 9 lines
tools/perf/builtin-stat.c +1
-0@@ -359,6 +359,7 @@ static void process_interval(void)clock_gettime (CLOCK_MONOTONIC , &ts);diff_timespec (&rs, &ts, &ref_time); +perf_stat__reset_rt_stat ();
could you use just existing:
perf_stat__reset_shadow_per_stat(&rt_stat)
or perhaps event perf_stat__reset_shadow_stats, given that we init walltime_nsecs_stats just few lines below
jirka
Quoted 3 lines
tools/perf/builtin-stat.c +0
-0read_counters (&rs);if (STAT_RECORD ) {
SNIP
Hi Jiri,
Quoted 16 lines · Jiri Olsa
On Fri, Apr 17, 2020 at 08:51:54AM +0800, Jin Yao wrote:
SNIP
tools/perf/builtin-stat.c static void process_interval(void) · line 362
Quoted 9 lines
tools/perf/builtin-stat.c +1
-0@@ -359,6 +359,7 @@ static void process_interval(void)clock_gettime (CLOCK_MONOTONIC , &ts);diff_timespec (&rs, &ts, &ref_time); +perf_stat__reset_rt_stat ();could you use just existing:
Quoted 7 lines
perf_stat__reset_shadow_per_stat(&rt_stat)or perhaps event
perf_stat__reset_shadow_stats, given that we initwalltime_nsecs_statsjust few lines belowjirka
Yes, I can use the existing function to reset rt_stat, I don't need to define a new one.
Posted v2 yet.
Thanks
Jin Yao
Quoted 6 lines
Quoted 3 lines
tools/perf/builtin-stat.c +0
-0read_counters (&rs);if (STAT_RECORD ) {SNIP