News for hackers. Stuff that matters.
Re:Intel Underestimates Error Bounds by 1.3 Quintillion by leoh on Thursday October 09, 2014 @06:22PM (#8435313)
I wonder what a great standard for this is. Incidentally, do TI or HP calculators give good results?
Re:Intel Underestimates Error Bounds by 1.3 Quintillion by StefanKarpinski on Thursday October 09, 2014 @08:18PM (#8435647)
Wolfram Alpha (and Mathematica) are generally completely accurate at the cost of being less efficient. I use those to check most things I want to verify.
Re:Intel Underestimates Error Bounds by 1.3 Quintillion by fnordfnordfnord on Friday October 10, 2014 @08:04AM (#8437279)
Nowhere near double precision.