Re: Big integer atoms
- Posted by "Igor Kachan" <kinz at peterlink.ru> Jul 12, 2004
- 523 views
Hello Christian, You wrote: ---------- > From: Christian Cuvier <Christian.CUVIER at agriculture.gouv.fr> > To: EUforum at topica.com > Subject: Re: Big integer atoms > Sent: 12 jul 2004 y. 13:40 > > > This "up to about 15 decimal digits" seems to be not > > enough concrete thing for some cases. > > > > Couldn't someone tell me about more precise > > bounds of these "larger integer values"? > > > > Thanks! > > > > Regards, > > Igor Kachan > > kinz at peterlink.ru > > > > When using doubles to store numbers (that's what Eu atoms physically are), 53 > out of the 64 available bits hold numerical info. The remainder is > for (biased) exponent and sign. This holds true on both Intel and m68k chips. > > 2^53 uses 15 or 16 decimal places to be written. This means that > integer computations with 15 decimal digits can still be done without > precision loss using atoms. At 16, you may encounter rounding off errors, and > there will be some at 17 and beyond quite definitely. I say "may" because of > the implied leading 1 in the mantissa, which gives you a little more leeway. > > Extended floats are more hardware specific, so that Eu can't easily > support them. And it would extend the no accuracy loss range to 18 digits > only. Thanks for the additional information. Good Luck! > Regards. > CChris Regards, Igor Kachan kinz at peterlink.ru