> Is there a document somewhere that describes the philosophy and how one makes > the choice? > My APO is applying an FIR to correct out-of-bounds frequency response in a > microphone. An EFX seemed like the right place for it. There's no white paper at this time, but the general philosophy is that the EFX and the hardware act as a unit to provide the absolute minimum amount necessary for basic functionality; the default location for processing is the MFX; and processing that absolutely has to act on individual streams goes in the SFX. RAW mode streams bypass the SFX and the MFX but not the EFX. Setting the "disable effects" switch effectively makes all streams RAW. There is a "verify audio effects discovery" test in the Hardware Compatibility Kit which has some sanity checks on what kind of processing can still be enabled with "disable effects" in place.