so i guess the point is... long double gives you more accuracy than double. but neither is "good enough" when you really need a LOT of bits.  :)


On 12/23/05, Rafael 'Dido' Sevilla <[EMAIL PROTECTED]> wrote:
Orlando Andico wrote:
> it's that "64 bits of mantissa" which will bite your ass in the end. of
> what use is an exponent of 4932 when there are lots of zeroes there?

The exponent is a side issue. The 19 digit accuracy is what's generally
more useful.  Eleven bits is a lot in numerical analysis, and that could
mean the difference between a catastrophic failure and correct
calculation in a sensitive numerical algorithm.  It's the sort of thing
people engaged in serious numerical analysis work tend to consider from
time to time.

_________________________________________________
Philippine Linux Users' Group (PLUG) Mailing List
[email protected] (#PLUG @ irc.free.net.ph)
Read the Guidelines: http://linux.org.ph/lists
Searchable Archives: http://archives.free.net.ph

Reply via email to