Hi,
I have 4 questions:
- In legacy pass manager, we have following sequence of code in PassManagerBuilder.cpp
PM.add( some_function_pass_1() )
PM.add( some_function_pass_2() )
why is it that we add function pass to a module pass manager? And does these 2 statements create 2 function pass managers ?
- In new pass manager, the above code would look something like this.
FunctionPassManager FPM;
FPM.add( some_function_pass_1() )
FPM.add( some_function_pass_2() )
MPM.add( std::move(FPM) )
Now is this code exactly identical to code in legacy pass manager shown above or how is it different?
-
Can adaptors be used in custom pass to change the schedule of pipeline?
-
Can adaptors be used to decide on the sequence of passes to be run such that sequence yields better results? By deciding I mean searching/trying out different sequences(like we search for good solutions in machine learning)
Regards
Sushant