The method Profiler Framework calculate the memory seems less accurate than the method from debugserver.

[YY-Wei]

The method Profiler Framework calculate the memory (https://github.com/wix/DetoxInstruments/blob/master/Profiler/DTXProfiler/PerformanceProfiling/DTXPerformanceSampler.m#L210)

result == KERN_SUCCESS ? task_vm_info.internal + task_vm_info.compressed : 0;
It doesn't seem that precise。

The memory profile method from debugserver seems more accurate.

anonymous = vm_info.internal + vm_info.compressed - vm_info.purgeable_volatile_pmap;

The value of anonymous is the same as the value of memory usage in Xcode Debug.
(https://github.com/llvm-mirror/lldb/blob/master/tools/debugserver/source/MacOSX/MachVMMemory.cpp#L144)

[LeoNatan]

Very interesting, thank you!
I will test and make the change soon.