> No, the x86 generates a trap for all divide overflows. If you had access to the errata sheet for one of the chips in common use, you might not be nearly so confident that implementations of the x86 architecture meet the specs. I personally know of many counter examples :-(