floating point inaccuracies

Thread view:

Enable EMail Alerts

Start New Thread

Thread rating:

Peteroid - 24 Mar 2005 01:28 GMT

Try putting this in your code:

double x(.008) ; // 8/1000
double y(.064) ; // 64/1000
double z(x+y) ; // 72/1000 ?

You'd think that z = .072. Nope, z = 0.072000000000000008. That seems REALLY
inaccurate to me. Is there a floating-point package (3rd party) that boasts
being more accurate AND is known to be reliable, simple to use, and
hopefully somewhat cheap? Thanx in advance for any help...! : )

Reply to this Message

Carl Daniel [VC++ MVP] - 24 Mar 2005 01:38 GMT

> Try putting this in your code:
>
[quoted text clipped - 6 lines]
> that boasts being more accurate AND is known to be reliable, simple to
> use, and hopefully somewhat cheap? Thanx in advance for any help...! : )

You need to learn what to expect from binary floating point. If you need
precise decimal accuracy (e.g. for financial work), you need to use a
fixed-point class, such as the System.Decimal type in .NET (which is a
wrapper over the OLE Automation Decimal type).

See http://docs-pdf.sun.com/800-7895/800-7895.pdf for lots of information on
what you can really expect from binary floating point math.

-cd

Reply to this Message

Free Magazines

Get these publications absolutely FREE for up to 12 months. There are no hidden fees and no obligation. Simply choose a title, complete the application form and submit it. Read more ...