Searching \ for 'Copy of: Re: 24 bit FLOATING POINT ROUTINES' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=bit+floating+point
Search entire site for: 'Copy of: Re: 24 bit FLOATING POINT ROUTINES'.

Truncated match.
PICList Thread
'Copy of: Re: 24 bit FLOATING POINT ROUTINES'
1995\08\28@121307 by David Bennett

picon face
Hello

I am using V1.18 of the 16c64 24 bit f.p. routines and have so far found 2
problems:-

1. The INT2424 routine dosn't seem to work for the case where the baised
exponent is h'90'

2. FPA24 has a problem which is showing up in a addition/divide routine I'm
using to give an average of several numbers. I'm not sure why, but for example,
adding together:-
       h'8c7f7f' + h'8c7f7f' gives h'8d7f7f'
which is correct, but
       h'8c7f80' + h'8c7f80' gives 8c7f00'
which is wrong.

I've cured problem 1. but still need to look at 2. - has anyone found the same
problem?

The 24 bits routines give reasonable accuracy in my application for a 5 digit
indicator - displayed numbers using all 5 digits are sometimes rounded by the
LSD or so - using the 32 bit versions would cure this but take more memory -
it's a hard life:-)

Hope this helps.

David Bennett

More... (looser matching)
- Last day of these posts
- In 1995 , 1996 only
- Today
- New search...