I am interested in using NEON & VFP pipeline and have the following questions,

1. I would like to optimize a code for NEON & VFP and hence would like to know best compiler options that will help to 
come up with optimized code(i.e. compiler options during development).

2. After optimization is done, to run the code on NEON & VFP what should be the best compiler options(i.e. Compiler 
option that will be used permanently).

3. Can I build a binary to use both NEON and VFP3 hardwares or I can use either VFP3/NEON ?
   i. If yes can I build the binary to use NEON pipe line for floating point operations and VFP hardware for double 
precision operations.