Author Topic: How to debug application driven CPU spike?  (Read 863 times)

Dariusz Piatkowski

  • Sr. Member
  • ****
  • Thank You
  • -Given: 13
  • -Receive: 8
  • Posts: 398
    • View Profile
How to debug application driven CPU spike?
« on: June 07, 2016, 03:54:39 pm »
I have a somewhat of a weird situation here.

While using the PUMonitor application I find that when I boot with the full 6 cores of my CPU enabled the app starts showing a consistent spike across all cores...usually about 11-15%. This does not happen if I boot with just 5 cores enabled.

Also what is interesting is that if I boot with 6 cores up (which is my normal mode of booting) and disable all but the single core the various CPU meteres I use (CPUMon, TOP or the WarpCenter built-in monitor) show a steady 50-60% spike that just continues and is all attributed to the PUMon utility app.

Sooo...the question is: given that the PUMon author has released the source code and that I have the OS/2 toolkit as well as the VAC++/IBM C/C++ toolsets installed (along with the debugger functionality) how would I take a stab at understanding where the CPU spike comes from?

Literally I am thinking of re-compiling the app on my machine and then invoking a DEBUG run...but I simply have no idea how to capture what in the app is causing the CPU spike. Looking for suggestions...

I've attached a screenshot of the CPU spikes and steady 50-60% utilization "flatline" I mentioned above.

Greg Pringle

  • Jr. Member
  • **
  • Thank You
  • -Given: 4
  • -Receive: 2
  • Posts: 97
    • View Profile
Re: How to debug application driven CPU spike?
« Reply #1 on: June 07, 2016, 08:11:41 pm »
Clearly a bug in the CPU monitor. I have seen this on many os/2 smp monitors. some will even hang the system.