IgProf is both a performance profiler and memory allocation profiler we will use at this school. This exercise will introduce you to the basic mechanics of using IgProf. Later exercises will go into much more detail, so don’t worry about understanding everything at this point; the main goal here is to learn to use the tools.
Basic profiling environment
To make sure you have technically everything set up right, let’s first make
sure you can create a dummy profile for simply running ls.
-
Make sure your environment is correct:
igprof -h # should print simple help message -
Create your very first igprof report on
ls:igprof -pp -z -o ls-test.gz -- ls -laFR /usr/lib >/dev/null igprof-analyse --sqlite -d -v -g ls-test.gz | sqlite3 ~/public_html/cgi-bin/data/ls-test.sql3 -
View the report at http://esc-gw.pd.infn.it:61100/~studentNM/cgi-bin/igprof-navigator.py/ls-test/
Basic performance profile
This exercise will run the program from the previous exercise on cputicks
using igprof statistical performance profiler.
-
Make sure your environment is correct:
igprof -h # should print simple help message -
Run the test program under igprof performance profiler:
cd esc15/hands-on/basic igprof -pp -z -o cputicks.gz ./cputicks 100000 -
Generate a simple text profile output of run time:
igprof-analyse -d -v -g cputicks.gz > cputicks.res less -SX cputicks.resYou should see an output like this:
Counter: PERF_TICKS ---------------------------------------------------------------------- Flat profile (cumulative >= 1%) % total Total Function 100.0 0.28 <spontaneous> [1] 100.0 0.28 _start [2] 100.0 0.28 __libc_start_main [3] 100.0 0.28 main [4] 97.8 0.27 test1(int) [5] <... and so on ...> -
Generate the same profile in web-viewable database, view the profile at http://esc-gw.pd.infn.it:61100/~studentNM/cgi-bin/igprof-navigator.py/cputicks/ :
rm -f ~/public_html/cgi-bin/data/cputicks.sql3 igprof-analyse --sqlite -d -v -g cputicks.gz | sqlite3 ~/public_html/cgi-bin/data/cputicks.sql3