[PATCH] perf timechart: explain the SVG output
From: Elazar Leibovich
Date: Mon Jun 01 2020 - 13:10:33 EST
It is not immediately clear what the SVG output means
and how is it supposed to be used.
Add links to example SVGs, and to annotated images
explaining what each part in the SVG means.
Signed-off-by: Elazar Leibovich <elazarl@xxxxxxxxx>
tools/perf/Documentation/perf-timechart.txt | 22 +++++++++++++++++++++
1 file changed, 22 insertions(+)
diff --git a/tools/perf/Documentation/perf-timechart.txt b/tools/perf/Documentation/perf-timechart.txt
index ef0c7565bd5c..460f3bd65789 100644
@@ -12,6 +12,13 @@ SYNOPSIS
+The perf-timechart tool will visualize what the system did in a certain timespan.
+Looking at 'perf timechart' output will allow you to see at a glance:
+ How much time was a certain process scheduled, when was it waiting for IO?
+ How many CPUs did the process use?
+ Did my process move between CPUs?
There are two variants of perf timechart:
'perf timechart record <command>' to record the system level events
@@ -30,6 +37,15 @@ There are two variants of perf timechart:
There are also poll bars which show how much time application spent
in poll/epoll/select syscalls.
+See also annotated example of perf-timechart report:
+Or annotated example of perf-timechart record -I:
@@ -123,6 +139,12 @@ Record system-wide IO events:
$ perf timechart
+Example SVGs for the wrk HTTP benchmark testing python3 -m http.server
+can be found here: