I have a DIMM with a bad bit which #memtest (86+ v 5.01) reports at 0045de6fa98, but it gives me #badram pattern of badram=0xdde6fa98,0xfffffffc ... Am I right to be concerned about the discrepancy in the upper bits between the reported address and the badram line? This is an i7 960 with 18GiB RAM in 6 DIMMs and (of course) will be running a 64 bit kernel. @catoutofbed