... is very often questionable. If you don't want to spend too much of precious FLASH (not to say even more precious RAM for "local variables" and stack, and/or other resources such as registers) space on them, they inevitably end up with very limited functionality. As an example, said monitors usually provide very little memory/performance ratio beyond basic "playing" with the chips, and in real application the developer tends to rewrite most of the routines in a more appropriate/specialized form anyway.

Whether this applies to your application(s) or not, only you can tell.

However, on some derivatives (but not SiLabs', apparently) it is a necessity - or convenience - to provide a set of functions on a separate piece of memory, whether FLASH or ROM (in the latter case, those functions are "built in" by factory). Often this is the case for in-application FLASH programming - as FLASH is non-readable during programming, in those architectures which don't "freeze" program execution during programming the code which performs the programing must be located outside the programmed FLASH. This leads straighforwardly to a IAP/ISP set of code in a dedicated RAM. You can perhaps have a look at how these work, for inspiration (appote AN461 of Philips/NXP, datasheet of Atmel's AT89C51RD2).

JW