Essential facts about floating point calculations

Re:Essential facts about floating point calculations by ucho on Monday July 06, 2015 @05:12AM (#9837771)

> Just consider that with 64-bit floats 0.3 + 0.6 = 8.9999999999999991.

> It really took no effort at all to get a precision problem!

That sure looks like serious precision problem :)

[ Parent | on HN ]

Re:Essential facts about floating point calculations by mortoray on Monday July 06, 2015 @05:40AM (#9837861)

I suppose in hindsight my example doesn't look as disturbing as it's meant to. :)

It's hard to show concise examples that truly show how this little bit of error eventually compounds and destabilizes a calculation.

[ Parent | on HN ]

Re:Essential facts about floating point calculations by jobigoud on Monday July 06, 2015 @05:56AM (#9837922)

I think he is referring to the fact that the result should be around 0.8999, not 8.9999.

[ Parent | on HN ]

Re:Essential facts about floating point calculations by jacquesm on Monday July 06, 2015 @06:55AM (#9838136)

It looks plenty disturbing to me. I don't even think it should be qualified as a 'precision problem' it's more like a 'total failure'.

[ Parent | on HN ]