The Saga of The Floating Point Flaw continues apace at Intel Corp and, according to another story on the Electronic Times Online, the problem may well be worse than we were first led to believe. Robert Collins, the man who disclosed details of the the flaw in the first place on his Intel Secrets Web site, originally assessed the problem as being one where the wrong type of error message was being flagged during overflow conditions that involved floating point addition. He has since had his diagnosis corrected by one Martin Atkinson-Barr, a self-employed physicist and self-styled expert on numerical analysis software from California, who says that rather than the wrong type of error being flagged you won’t get an error indication at all. Despite its promises to respond with a report by Friday, Intel still hadn’t published anything to shed any more light on the problem by the time Computergram went to press.
