Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1304865
| From | Arnaldo Carvalho de Melo <acme@kernel.org> |
|---|---|
| Newsgroups | linux.kernel |
| Subject | [PATCH 30/59] perf script: Add stat-cpi.py script |
| Date | 2016-01-08 19:30 +0100 |
| Message-ID | <qOJXt-29h-43@gated-at.bofh.it> (permalink) |
| References | <qOJE5-21e-1@gated-at.bofh.it> |
| Organization | linux.* mail to news gateway |
From: Jiri Olsa <jolsa@kernel.org>
Adding stat-cpi.py as an example of how to do stat scripting.
It computes the CPI metrics from cycles and instructions events.
The CPI is based performance metric showing the Cycles Per Instructions
ratio, which helps to identify cycles-hungry code.
Following stat record/report/script combinations could be used:
- get CPI for given workload
$ perf stat -e cycles,instructions record ls
SNIP
Performance counter stats for 'ls':
2,904,431 cycles
3,346,878 instructions # 1.15 insns per cycle
0.001782686 seconds time elapsed
$ perf script -s ./scripts/python/stat-cpi.py
0.001783: cpu -1, thread -1 -> cpi 0.867803 (2904431/3346878)
$ perf stat -e cycles,instructions record ls | perf script -s ./scripts/python/stat-cpi.py
SNIP
0.001730: cpu -1, thread -1 -> cpi 0.869026 (2928292/3369627)
- get CPI systemwide:
$ perf stat -e cycles,instructions -a -I 1000 record sleep 3
# time counts unit events
1.000158618 594,274,711 cycles (100.00%)
1.000158618 441,898,250 instructions
2.000350973 567,649,705 cycles (100.00%)
2.000350973 432,669,206 instructions
3.000559210 561,940,430 cycles (100.00%)
3.000559210 420,403,465 instructions
3.000670798 780,105 cycles (100.00%)
3.000670798 326,516 instructions
$ perf script -s ./scripts/python/stat-cpi.py
1.000159: cpu -1, thread -1 -> cpi 1.344823 (594274711/441898250)
2.000351: cpu -1, thread -1 -> cpi 1.311972 (567649705/432669206)
3.000559: cpu -1, thread -1 -> cpi 1.336669 (561940430/420403465)
3.000671: cpu -1, thread -1 -> cpi 2.389178 (780105/326516)
$ perf stat -e cycles,instructions -a -I 1000 record sleep 3 | perf script -s ./scripts/python/stat-cpi.py
1.000202: cpu -1, thread -1 -> cpi 1.035091 (940778881/908885530)
2.000392: cpu -1, thread -1 -> cpi 1.442600 (627493992/434974455)
3.000545: cpu -1, thread -1 -> cpi 1.353612 (741463930/547766890)
3.000622: cpu -1, thread -1 -> cpi 2.642110 (784083/296764)
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Tested-by: Kan Liang <kan.liang@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1452077397-31958-4-git-send-email-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/scripts/python/stat-cpi.py | 77 +++++++++++++++++++++++++++++++++++
1 file changed, 77 insertions(+)
create mode 100644 tools/perf/scripts/python/stat-cpi.py
diff --git a/tools/perf/scripts/python/stat-cpi.py b/tools/perf/scripts/python/stat-cpi.py
new file mode 100644
index 000000000000..8b60f343dd07
--- /dev/null
+++ b/tools/perf/scripts/python/stat-cpi.py
@@ -0,0 +1,77 @@
+#!/usr/bin/env python
+
+data = {}
+times = []
+threads = []
+cpus = []
+
+def get_key(time, event, cpu, thread):
+ return "%d-%s-%d-%d" % (time, event, cpu, thread)
+
+def store_key(time, cpu, thread):
+ if (time not in times):
+ times.append(time)
+
+ if (cpu not in cpus):
+ cpus.append(cpu)
+
+ if (thread not in threads):
+ threads.append(thread)
+
+def store(time, event, cpu, thread, val, ena, run):
+ #print "event %s cpu %d, thread %d, time %d, val %d, ena %d, run %d" % \
+ # (event, cpu, thread, time, val, ena, run)
+
+ store_key(time, cpu, thread)
+ key = get_key(time, event, cpu, thread)
+ data[key] = [ val, ena, run]
+
+def get(time, event, cpu, thread):
+ key = get_key(time, event, cpu, thread)
+ return data[key][0]
+
+def stat__cycles_k(cpu, thread, time, val, ena, run):
+ store(time, "cycles", cpu, thread, val, ena, run);
+
+def stat__instructions_k(cpu, thread, time, val, ena, run):
+ store(time, "instructions", cpu, thread, val, ena, run);
+
+def stat__cycles_u(cpu, thread, time, val, ena, run):
+ store(time, "cycles", cpu, thread, val, ena, run);
+
+def stat__instructions_u(cpu, thread, time, val, ena, run):
+ store(time, "instructions", cpu, thread, val, ena, run);
+
+def stat__cycles(cpu, thread, time, val, ena, run):
+ store(time, "cycles", cpu, thread, val, ena, run);
+
+def stat__instructions(cpu, thread, time, val, ena, run):
+ store(time, "instructions", cpu, thread, val, ena, run);
+
+def stat__interval(time):
+ for cpu in cpus:
+ for thread in threads:
+ cyc = get(time, "cycles", cpu, thread)
+ ins = get(time, "instructions", cpu, thread)
+ cpi = 0
+
+ if ins != 0:
+ cpi = cyc/float(ins)
+
+ print "%15f: cpu %d, thread %d -> cpi %f (%d/%d)" % (time/(float(1000000000)), cpu, thread, cpi, cyc, ins)
+
+def trace_end():
+ pass
+# XXX trace_end callback could be used as an alternative place
+# to compute same values as in the script above:
+#
+# for time in times:
+# for cpu in cpus:
+# for thread in threads:
+# cyc = get(time, "cycles", cpu, thread)
+# ins = get(time, "instructions", cpu, thread)
+#
+# if ins != 0:
+# cpi = cyc/float(ins)
+#
+# print "time %.9f, cpu %d, thread %d -> cpi %f" % (time/(float(1000000000)), cpu, thread, cpi)
--
2.1.0
Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
[GIT PULL 00/59] perf/core improvements and fixes Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 24/59] perf script: Process stat config event Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 40/59] tools lib: Sync tools/lib/find_bit.c with the kernel Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 49/59] perf tools: Add overhead/overhead_children keys defaults via string Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 39/59] tools lib: Move find_next_bit.c to tools/lib/ Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 42/59] perf top: Decay periods in callchains Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 15/59] tools build feature: Fix feature_check_display_code typo Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 23/59] perf script: Process cpu/threads maps Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 43/59] perf report: Change default to use event group view Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 05/59] perf tools: Pass evlist to setup_sorting() Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 10/59] perf tools: Support shortcuts for events in dynamic sort keys Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 13/59] perf tools: Add 'trace_fields' dynamic sort key Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 59/59] perf evlist: Add --trace-fields option to show trace fields Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 52/59] perf report: Show random usage tip on the help line Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 48/59] perf tools: Remove list entry from struct sort_entry Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 14/59] perf tools: Make 'trace' or 'trace_fields' sort key default for tracepoint events Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 25/59] perf script: Add process_stat/process_stat_interval scripting interface Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 41/59] tools lib: Move bitmap.[ch] from tools/perf/ to tools/{lib,include}/ Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 02/59] perf hist: Save raw_data/size for tracepoint events Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 08/59] perf tools: Add 'trace' sort key Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 09/59] perf report/top: Add --raw-trace option Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 45/59] perf tools: Add missing headers in perf's MANIFEST Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:10 +0100
[PATCH 57/59] perf libdw: Check for mmaps also in MAP__VARIABLE tree Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 56/59] perf unwind: Check for mmaps also in MAP__VARIABLE tree Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 53/59] perf evlist: Make perf_evlist__open() open evsels with their cpus and threads (like perf record does) Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 58/59] perf record: Store data mmaps for dwarf unwind Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 47/59] perf tools: Include all tools/lib directory for tags/cscope/TAGS targets Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 54/59] perf evlist: Remove perf_evlist__(enable|disable)_event functions Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 51/59] perf hists: Export a couple of hist functions Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 12/59] perf tools: Skip dynamic fields not defined for current event Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 03/59] tools lib traceevent: Factor out and export print_event_field[s]() Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 38/59] perf tests: Give a bit more information on the CQM test failure path Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 37/59] perf tests: No need to set attr.sample_freq for tracking !PERF_RECORD_SAMPLE Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 11/59] perf tools: Support '<event>.*' dynamic sort key Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 04/59] perf top: Create the evlist sooner Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 46/59] perf script: Align event name properly Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 50/59] perf diff: Use perf_hpp__register_sort_field interface Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 18/59] tools build feature: Use value assignment form for FEATURE-DUMP file Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 35/59] perf test: No need for setting attr.sample_freq on the RECORD test Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 33/59] perf evlist: Introduce perf_evlist__new_dummy constructor Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 36/59] perf python: Add missing files to binding link list Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 55/59] perf unwind: Use find_map function in access_dso_mem Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:20 +0100
[PATCH 19/59] perf build: Use FEATURE-DUMP in bpf subproject Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 28/59] perf cpumap: Fix cpu conversion in cpu_map__from_entries Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 29/59] perf script: Display stat events by default Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 07/59] perf tools: Try to show pretty printed output for dynamic sort keys Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 16/59] tools build feature: Move dwarf post unwind choice output into perf Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 20/59] perf tools: Add all matching dynamic sort keys for field name Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 26/59] perf script: Add stat default handlers Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 44/59] perf tools: Do not show trace command if it's not compiled in Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 27/59] perf script: Add python support for stat events Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 21/59] perf report: Add documentation for dynamic sort keys Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 30/59] perf script: Add stat-cpi.py script Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 34/59] perf test: Use "dummy" events in the PERF_RECORD_ test Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 17/59] tools build feature: Introduce feature_assign macro Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 31/59] perf pmu: fix alias->snapshot missing initialization bug Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 32/59] perf tests: No need to set attr.sample_freq in the perf time to TSC test Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 01/59] perf hist: Pass struct sample to __hists__add_entry() Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 06/59] perf tools: Add dynamic sort key for tracepoint events Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
[PATCH 22/59] perf stat record: Keep sample_type 0 for pipe session Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-01-08 19:30 +0100
Re: [GIT PULL 00/59] perf/core improvements and fixes Ingo Molnar <mingo@kernel.org> - 2016-01-09 17:30 +0100
csiph-web