Hi,
This MR includes various improvements in PerfMonMT. Here is a list of changes:
cpu time per event
and add call count
metric in the component level monitoring of parallel steps. call count
simply shows that how many times a component is called in the event loop.smaps
file for every component, we have changed the way we capture the memory metrics in component level monitoring of serial steps. We read only vmem
and malloc
. vmem
is captured from statm
file of /proc
and malloc
is captured from mallinfo
function.Many thanks,
Hasan