I know that the pinout of the chip on the PCB is not the same as in the 1979 doc; at the very least Vdd was moved from pin 12 to pin 2 and pins 2-11 moved down one spot. And from looking at the die, I think other pins are different, too.
When I decapped the chip some plastic stuck to the center of the die, but the edge of the die is clear, so you can see the bonding pads. I uploaded a 150MB Gimp file: www.seanriddle.com/wildfire.xcf
Pins 1-20 look like they match the pinout (with pin 12 moved up to pin 2), starting at the top center of the die and going counter-clockwise. A3-A0 at bottom left are beefier than A12-A4 because they can drive 25mA instead of 5mA. And going clockwise from top center, pins 40-36 along the top look like the rest of the data bus. But pin 21 looks like it is connected to pin 1 (and is a NC on the Wildfire PCB). There are only 38 pads on the die, so at least 3 of the chip pins are no-connects. Pin 29 on the lower right side looks like Vgg, but it's hard for me to tell what the other pins are.
With a transistor on each side of the speaker, I wonder if one supplies pulses to make various frequencies and the other just turns the sound on and off. Maybe the K inputs were replaced by some frequency generating circuitry? Pins 30 and 33 (K1 and K8) are no-connects on the PCB, 31 is connected to a cap to Vss and 32 to a transistor, like pins C and F in the patent schematic. I wonder if the cap on C sets the frequency that's output on F.
