The Atmel parts give you internal EEProm for NVRAM, and the code space has small block sizes (128 and 1 byte) for programming. You do not need to worry about code location and blocks when programming the code using FLIP. You do not even have to do chip erase to re-program (although it is slower) !

The Philips parts give you some reasonable sized (8K) blocks for data storage, BUT they give them at 0000H - makes programming harder because this includes the interrupt vectors. The other blocks are big.

Atmel gives gou many ways to recover from improper boot vector, Philips gives none.