THE TRANSFER ROUTINE TAKES SIX ARGUMENTS -- NONE IS A DESTINATION.

Breakpoint at file 0x1253AE (after all six argument loads in 0x12538C):

  ARGS a0=00916B50  a1=00000180  a2=00000000
       a3=00000004  a4=00000000  a7=00000000

  a0 = 0x00916B50   main-RAM bit address -> the SOURCE pixels
  a1 = 0x180 (384)  a count -- note the engine's output port 0x5600200 is read
                    exactly 384 times, so a1 is almost certainly the number of
                    words/units to move
  a2 = 0
  a3 = 4
  a4 = 0
  a7 = 0            the word handed to the data register

There is still NO destination: no VRAM address (>= 0x04000000) in any argument,
in any register write, or in the 154-word table.  The engine is expected to place
the pixels using state programmed before the call, and that state is not a plain
address in anything observable.

CUMULATIVE RESULT
  The protocol stack is complete and verified end to end:
    DEC 2.3 client (built from DEC's source)
      -> XIE session, photomap, transport, PutTile            (verified import)
      -> Export accepted, DDX gates passed                    (export completes)
      -> source pixels streamed into the accelerator FIFO     (0xFFFF / 0x0000)
      -> transfer routine and all six arguments identified
  The single missing link is the engine's placement programming.

STATE
  * NO client-chosen XIE image has ever been displayed;
  * framebuffer unchanged; white vs black differ only by TM figures + cursor;
  * model opt-in (VT1200_ACCEL=1); default configuration unchanged;
  * xshow (core X) remains the only viewer that has shown an image.
