There seems to be a difference between the dirty size captured by VM Tracker and the dirty size captured by Xcode Debugger mode or Memory Graph.
I found out that resident memory size in VM Tracker is near 250MiB and the used memory metric in Xcode Debugger is around 730MB.
So I captured Memory graph and compared metrics between VM Tracker and Memory Graph.
First thing I found is that Xcode Debugger's Memory usage metric is actually memory footprint(Dirty + uncompressed size of Dirty Compressed).
(https://i.stack.imgur.com/b1xce.jpg)
And then I found out that Dirty size or Resident size is very different between VM Tracker and Memory Graph.
For example, the dirty size of MALLOC_LARGE was 125M in the Memory Graph, but it was around 39MiB in VM Tracker.
I'm curious why the results of the two profilers are different and which tool's results I should trust more.
EDIT:
These are different scenarios I tried but outcome was all same.
#1 profile same process
- Record VM Tracker for process 1
- at the sometime, attach Xcode debugger to same process 1
#2 Profile same build, different process instance
- Record VM Tracker for process 1, and finish recoding, terminate process 1.
- Run same build-process 2
- Attach Xcode debugger to process 2(same build, different instance so it won't affected by VM tracker)