News for hackers. Stuff that matters.
Re:Intel Underestimates Error Bounds by 1.3 Quintillion by pedrow on Thursday October 09, 2014 @04:48PM (#8434951)
Possible solution: revive the Indiana Pi Bill[0] (assuming 3.2 has an exact representation in binary?)

[0]:http://en.wikipedia.org/wiki/Indiana_Pi_Bill

Re:Intel Underestimates Error Bounds by 1.3 Quintillion by leni536 on Thursday October 09, 2014 @05:00PM (#8435004)
assuming 3.2 has an exact representation in binary?

It does not have an exact representation in binary.

Re:Intel Underestimates Error Bounds by 1.3 Quintillion by thaumasiotes on Thursday October 09, 2014 @11:29PM (#8436130)
Easy solution: amend the value to 3.25, or 3.125