New findings:
- the "status register" test (#4) pokes the GDC command port $57 circa 28 times with 0x22 (WDAT ?), then queries the status port $56. It then expects a
FIFO_FULL response (instead of EMPTY..as it is now).
Is it the flags or the timing?
- later tests of the GDC diagnostic disc often invoke
RDAT with mod = 02 (unimplemented ?).
On the positive side, the scroll test (#2) looks good, so we are at 23 % now ;-)