[maemo-developers] Profiling on Nokia N810

[From: Bruno]

Hello everybody !

 I've been trying to profile my program on the nokia N810 for some
 times now, and I'm not able to get good results. I installed the
 oprofile package and the oprofile modified kernel, and then tried to
 run gprof.

 I tried 2 ways to get my profiling information (my program is called
 src ... not really explicit I know ! ) :

 compile with normal compiler paramaters, no -pg. Then :

 Nokia-N810:~# opcontrol --init
 Nokia-N810:~# opcontrol --no-vmlinux
 Nokia-N810:~# opcontrol -e=CPU_CYCLES:100000
 Nokia-N810:~# opcontrol --start
 Nokia-N810:~# ./src
 Nokia-N810:~# opcontrol --stop
 Nokia-N810:~# opgprof src
 Nokia-N810:~# gprof src > lala.txt
 gprof: gmon.out is file is missing call-graph data
 Nokia-N810:~# gprof -Q src > lala.txt

 and here is the text file I obtain :

 Flat profile:

 Each sample counts as 1 samples.
  %   cumulative   self              self     total
  time   samples   samples    calls  T1/call  T1/call  name
  26.86   9243.00  9243.00                             VjDetect_DetectFace
  14.55  14250.00  5007.00                             _init
  14.51  19243.00  4993.00
 detect_face(mainstruct*, _IplImage*)
  13.38  23847.00  4604.00
 VjDetect_SetWindowSizeInCascade
  11.80  27908.00  4061.00                             LBP(_IplImage*)


 As you see, I have the percentage of time each function take, but not
 the number of calls...

 So I tried something else :
 I compile with -pg option for linker and compiler.
 Then I use gprof the way I would do on a normal linux architecture.

 Nokia-N810:~# ./src
 Nokia-N810:~# gprof src > lala.txt

 And what I get is :

 Flat profile:

 Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total
  time   seconds   seconds    calls  Ts/call  Ts/call  name
 100.01      0.01     0.01                             LoadCascade
  0.00      0.01     0.00  2663842     0.00     0.00  CalcSum
  0.00      0.01     0.00    14708     0.00     0.00  CalcSqSum
  0.00      0.01     0.00    14708     0.00     0.00  VjDetect_DetectFace
  0.00      0.01     0.00       12     0.00     0.00
 VjDetect_SetWindowSizeInCascade
  0.00      0.01     0.00        9     0.00     0.00  cvSize(int, int)

 or :

 Flat profile:

 Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total
  time   seconds   seconds    calls  Ts/call  Ts/call  name
  50.01      0.01     0.01                             DoCoarseScan
  50.01      0.02     0.01                             LoadCascade
  0.00      0.02     0.00  6568635     0.00     0.00  CalcSum
  0.00      0.02     0.00    40447     0.00     0.00  CalcSqSum
  0.00      0.02     0.00    40447     0.00     0.00  VjDetect_DetectFace
  0.00      0.02     0.00       33     0.00     0.00
 VjDetect_SetWindowSizeInCascade
  0.00      0.02     0.00       30     0.00     0.00  cvSize(int, int)


 or even :

 Flat profile:

 Each sample counts as 0.01 seconds.
  no time accumulated

  %   cumulative   self              self     total
  time   seconds   seconds    calls  Ts/call  Ts/call  name
  0.00      0.00     0.00  5042653     0.00     0.00  CalcSum
  0.00      0.00     0.00    25739     0.00     0.00  CalcSqSum
  0.00      0.00     0.00    25739     0.00     0.00  VjDetect_DetectFace
  0.00      0.00     0.00       21     0.00     0.00
 VjDetect_SetWindowSizeInCascade
  0.00      0.00     0.00       18     0.00     0.00  cvSize(int, int)
  0.00      0.00     0.00        7     0.00     0.00  VjDetect_DetectAllFaces
  0.00      0.00     0.00        7     0.00     0.00  VjDetect_IntegralImage


 So now I have the number of calls, but no or weird times informations.
 For information I used the precompiled maemo diablo kernel for oprofile.
 Anybody got a similar problem ? Or a solution ?

 Thanks in advance, I'm starting to desesperate to profile this one !

 Best regards,

Bruno