[net.micro.cbm] Microsoft mathpack

fred@cbmvax.cbm.UUCP (Fred Bowen) (08/11/86)

First the good news-  the so-called 'update' ROMs for the C128 are finally
getting the green light for release to manufacturing.  How they will be made
available to the public is not yet certain.  Tune in next week for a list
of changes and (hopefully) instructions for acquiring these ROMs.  At the
same time I will be releasing the 1571 disk drive update ROM.

Now the bad news- I have been unable to devise an acceptable fix for a problem
with the math package.  First a little explanation.  There currently exists a
problem in the C64 mathpack (and other computers utilizing the Microsoft math
routines, such as 8032s, PETs, Apples) which has been known for some time now
as the 'double zero' bug.  It affects multiplies in which the multiplier has
two consecutive zero bytes in the FAC, resulting in an extra bit shift.  This
can be  seen  by  evaluating  (on a C64)  something  like  50*16777217  vs.
16777217*50.  They give different answers (the latter one is correct).  This
has been fixed on the C128 from the very beginning.  Unfortunately it appears
to have screwed up some unfathomable 'fudge' values in the EXP routine, such
that, for example, 2^15 now yields 32768.0001.  Barf.

So maybe someone can help me.  What I am looking for is a book.  I am told
it was written by one of the Microsoft gurus responsible for the mathpack
and does a fair job of explaining some of the 'unexplainable'.  I know neither
his name nor the title of the book.  Gad- I'm on a roll, aren't I??  Can
someone help me?
-- 
Fred Bowen			uucp:	{ihnp4|seismo|caip}!cbmvax!fred
				arpa:	cbmvax!fred@seismo.CSS.GOV
				tele:	215 431-9100

Commodore Electronics, Ltd.,  1200 Wilson Drive,  West Chester,  PA,  19380