I think it got something to do with your start vector. As your code may consist of more than 8 bytes. Try to modify the following part.

ORG 0H
ajmp loop

ORG 100H ;anything after 30H should
;be okay

loop: "your code"

Hope that helps !

Best Regards
Andy

P.S Maybe your delay routine can use
"DJNZ" instead of "cjne" which is
much more neater. But anyway, it's
just a SUGGESTION. Your code looks
fine :)