Last night I uploaded the Beige ROM chip contents to:
http://www.prismnet.com/~trag/Firmware/Beige/
The files Beige_C_U1.bin, Beige_C_U2.bin, Beige_C_U3.bin, and Beige_C_U4.bin are the files for the chips.
I also made four dumps of my Beige G3 and put them in that folder, but none of them match each other, and for reasons explained below, I believe that all four of them are garbage.
I pulled some code out of the chip files at address (you need a hex editor to see what I'm discussing here) 0x0000 0310:
U1:
..OY. . P"uN.d.ker..mHp.. C./
U3:
sn./@ .gR ......ne.. . Q .../
U2:
dr . ) . oSDECB....n U$ A #./
U4:
.f."" . .NnusicS....mHjH. #
Interleaving it by taking two bytes at a time from each chip, and reversing the order of those two bytes (characters), yields the following:
..nsrdf.YO/. ". . @ . . .g. ) ."P R .N.Nu..Sound...EDisk...BCScreen............Hm nHm.p .$UHj . QA .C ..# # /./././<
Which is not to be found anywhere in the extracts I made, but it is found starting at address 0x0C40 (0x0310 X 4) in the extract at the Mac ROM archive.
There's another big useful chunk of text at 0x1ACB in the Beige extract and it corresponds with text found in the chip files a little before 0x06B0, although I haven't done a full decode to be absolutely certain.
So, this suggests that the interleave/deinterleave method is the same for the Beige G3 as for the 9600, which is promising when it comes to deinterleaving the PEx extract.
Now, on to the PEx extracts. I have various ones on hand. I'll try to get those uploaded soon. The question will be how do they differ, and which one do I use? There are 3 MB and 4 MB versions, and I hope that all the 3 MB versions are identical to each other and all the 4 MB versions are identical to each other. That would simplify matters to choosing between those two, or using them both and seeing which one works.