Profiling - Revision history

T1t0 at 01:29, 23 August 2023

2023-08-23T01:29:15Z

← Older revision		Revision as of 01:29, 23 August 2023
Line 38:		Line 38:
	</pre>		</pre>

	Example of output of [http://~~github~~.com/campisano/stsm STSM] program:		Example of output of [http://gitlab.com/campisano/stsm STSM] program:

	<pre>		<pre>

2022-04-06T21:38:26Z

perf tool

← Older revision		Revision as of 21:38, 6 April 2022
Line 18:		Line 18:
	</pre>		</pre>

			or the more gdb specific
			<pre>
			-Og -ggdb3 -fno-omit-frame-pointer
			</pre>

	There are a lot of event that can be monitored, and they can be specified using the -e option. The default event is just 'cycles' (or 'cycles:pp' apparently), but the option accept multiple event types separated by a comma, like '-e cpu-clock,faults' for instance.		There are a lot of event that can be monitored, and they can be specified using the -e option. The default event is just 'cycles' (or 'cycles:pp' apparently), but the option accept multiple event types separated by a comma, like '-e cpu-clock,faults' for instance.

2021-03-14T00:44:17Z

References

← Older revision		Revision as of 00:44, 14 March 2021
Line 115:		Line 115:
	* [https://stackoverflow.com/a/3861464/846686 other tools]		* [https://stackoverflow.com/a/3861464/846686 other tools]

	[[Category:cpp]]		[[Category:c/cpp]]
	[[Category:debug]]		[[Category:debug]]
	[[Category:profiling]]		[[Category:profiling]]

2019-12-29T20:26:27Z

@@ Line 71: / Line 71: @@
 perf script | c++filt | gprof2dot.py -f perf | dot -Tpng -o output.png
 </pre>
 == valgrind tool ==
@@ Line 79: / Line 88: @@
 qcachegrind I100_Ooriginal_S26_FS10_FB05_MS0.grind
 </pre>
 = References =
 * [https://stackoverflow.com/a/3861464/846686 other tools]
 [[Category:cpp]]
 [[Category:debug]]
 [[Category:profiling]]

2019-12-29T20:20:02Z

perf tool

@@ Line 54: / Line 54: @@
 +   49.99%    49.99%  stsm     stsm                  [.] Segment::intersects
 +   25.22%     0.00%  stsm     stsm                  [.] STSM::detectBlocksOfAllSolidSequences
 </pre>

2019-12-29T16:31:13Z

References

← Older revision		Revision as of 16:31, 29 December 2019
Line 67:		Line 67:

	* [https://perf.wiki.kernel.org/index.php/Tutorial perf Tutorial]		* [https://perf.wiki.kernel.org/index.php/Tutorial perf Tutorial]
			* perf command manuals: [http://man7.org/linux/man-pages/man1/perf-record.1.html record] and [http://man7.org/linux/man-pages/man1/perf-report.1.html report]
	* [http://www.brendangregg.com/perf.html Linux perf Examples]		* [http://www.brendangregg.com/perf.html Linux perf Examples]
	* [https://stackoverflow.com/a/3861464/846686 other tools]		* [https://stackoverflow.com/a/3861464/846686 other tools]

2019-12-29T16:30:18Z

perf tool

← Older revision		Revision as of 16:30, 29 December 2019
Line 22:		Line 22:

	<pre>		<pre>
	perf record -e cycles -g --call-graph fp -- ./COMMAND ARG1 ARG2 ... ARGN		perf record -q -e cycles -g --call-graph fp -- ./COMMAND ARG1 ARG2 ... ARGN
	</pre>		</pre>

Line 28:		Line 28:

	<pre>		<pre>
	perf report -g graph -~~F +sample,~~period,comm,dso,symbol		perf report --stdio -g none --sort comm,dso # load per module
			perf report --stdio -g none # load per function
			perf report --stdio -g graph # call chains
			perf report -g graph -s period,comm,dso,symbol # TUI interface
	</pre>		</pre>

2019-12-29T07:50:14Z

References

← Older revision		Revision as of 07:50, 29 December 2019
Line 67:		Line 67:
	* [https://stackoverflow.com/a/3861464/846686 other tools]		* [https://stackoverflow.com/a/3861464/846686 other tools]
	* [http://rhaas.blogspot.com.br/2012/06/perf-good-bad-ugly.html tools considerations]		* [http://rhaas.blogspot.com.br/2012/06/perf-good-bad-ugly.html tools considerations]
	* <s>[https://stackoverflow.com/a/10958510/846686 perf example (~~out~~)]</s>, [https://web.archive.org/web/20160602190008/http://stackoverflow.com/questions/1777556/alternatives-to-gprof/10958510#10958510 webarchive link])		* <s>[https://stackoverflow.com/a/10958510/846686 perf example (offline)]</s>, [https://web.archive.org/web/20160602190008/http://stackoverflow.com/questions/1777556/alternatives-to-gprof/10958510#10958510 webarchive link]
	* [https://unix.stackexchange.com/a/385129/ simplest perf example]		* [https://unix.stackexchange.com/a/385129/ simplest perf example]

2019-12-29T07:49:46Z

← Older revision		Revision as of 07:49, 29 December 2019
Line 64:		Line 64:

	* [https://perf.wiki.kernel.org/index.php/Tutorial perf Tutorial]		* [https://perf.wiki.kernel.org/index.php/Tutorial perf Tutorial]
			* [http://www.brendangregg.com/perf.html Linux perf Examples]
	* [https://stackoverflow.com/a/3861464/846686 other tools]		* [https://stackoverflow.com/a/3861464/846686 other tools]
	* [http://rhaas.blogspot.com.br/2012/06/perf-good-bad-ugly.html tools considerations]		* [http://rhaas.blogspot.com.br/2012/06/perf-good-bad-ugly.html tools considerations]
	* [https://stackoverflow.com/a/10958510/846686 perf example (out, [https://web.archive.org/web/20160602190008/http://stackoverflow.com/questions/1777556/alternatives-to-gprof/10958510#10958510 webarchive link])]		* <s>[https://stackoverflow.com/a/10958510/846686 perf example (out)]</s>, [https://web.archive.org/web/20160602190008/http://stackoverflow.com/questions/1777556/alternatives-to-gprof/10958510#10958510 webarchive link])
	* [https://unix.stackexchange.com/a/385129/ simplest perf example]		* [https://unix.stackexchange.com/a/385129/ simplest perf example]

2019-12-29T07:48:28Z

perf tool

← Older revision		Revision as of 07:48, 29 December 2019
Line 2:		Line 2:

	== perf tool ==		== perf tool ==

			* Install

			<pre>
			apt-get install linux-perf
			</pre>

	* Produce profile data		* Produce profile data

	Compile with ~~'-ggdb' and~~ '-fno-omit-frame-pointer' options as CFLAGS. An alternative exists, using the '--call-graph dwarf' option, however it apparently generates a lot of more disk data. To know more about such option, take a look at the link http://www.brendangregg.com/perf.html#StackTraces and use 'perf help record' to show all the options.		Compile with '-fno-omit-frame-pointer' options as CFLAGS. An alternative exists, using the '--call-graph dwarf' option, however it apparently generates a lot of more disk data. To know more about such option, take a look at the link http://www.brendangregg.com/perf.html#StackTraces and use 'perf help record' to show all the options.

	For instance, use		For instance, use
	<pre>		<pre>
	-Og ~~-ggdb~~ -g3 -fno-omit-frame-pointer		-Og -g3 -fno-omit-frame-pointer
	</pre>		</pre>

Line 16:		Line 22:

	<pre>		<pre>
	~~perf record -g --call-graph fp -- ./your_app your_args~~		perf record -e cycles -g --call-graph fp -- ./COMMAND ARG1 ARG2 ... ARGN
	~~</pre>~~

	or

	~~<pre>~~
	perf record -e cycles -g --call-graph fp -- ./~~your_app your_args~~
	</pre>		</pre>

Line 28:		Line 28:

	<pre>		<pre>
	perf report -n -g graph		perf report -g graph -F +sample,period,comm,dso,symbol
	</pre>		</pre>