Does it only execute the loop once and get stuck for the first time at 2574H?
The 49H in the data dump is most likely the C9H for the RET. The next two byte are the return address from the stack. They could be 0A57H, 0AD7H, 8A57, 8AD7H or other permutations. As this routine is writing to the RAM it is quite possible it is trashing the stack and destroying the return address - could be a bad RAM chip or decoder. There are at least two different versions of the M100 motherboard one has 700mil RAM modules and the other 750mil. Not sure if there are any ROM differences or other hardware differences.
