Even FWs that measure identically flat sound different. Easily done with 2 clips *28 and *31 as well as Rockbox can easily be distinguished in an AB. I think the proper response is 'whatever' from here on out. Skamp is absolutely correct about output power and impedance. Not so much that the simplest measurements tell you everything.
Yeah we don't set the DAC quite the same as the sandisk firmware. The stereo separation is different in rockbox (better actually!), and the pitch is more accurate. I'm surprised that different sandisk firmwares sound different though. When I disassembled them it seemed like they changed very little between them. Actually, it seemed like sandisk was hesitant to change much of anything.