Having had a few minutes, AND an 8052 board, AND a DS89C450, admittedly not the exact same part, but OK for starters ...

I powered up the board with the DS89C450 in place, along with no program in flash and no external memory (I had to remove it).
The external ROM contained a monitor with which I was able to ensure that it didn't believe it had any RAM at 0000H. I then OR'd 0x81 to the PMR (0xC4). After that, it magically had RAM at 000 - 3FF, which I could verify at will with the monitor's fill and dump commands, each time getting back what I expected.

Now, this wasn't a rigorous memory test, but it certainly proves that setting bit 0 of SFR at 0xC4 makes the board behave very differently than when bit 0 is cleared. This certainly establishes one thing, and that's that THIS device enables its internal RAM block when that bit is set, and doesn't when it's not.

That's all I had time for, but I'll do a few other things when I put a DS89C420 in the board, ultimately putting the code Matthew Hay listed in program space and booting to that. It's a socketed part, so that'll be no trouble.

RE