Bruce Evans a0e34da09f Oops, the last-minute optimization in rev.1.8 wasn't a good idea. The
17+17+24 bit pi/2 must only be used when subtraction of the first 2
terms in it from the arg is exact.  This happens iff the the arg in
bits is one of the 2**17[-1] values on each side of (float)(pi/2).

Revert to the algorithm in rev.1.7 and only fix its threshold for using
the 3-term pi/2.  Use the threshold that maximizes the number of values
for which the 3-term pi/2 is used, subject to not changing the algorithm
for comparing with the threshold.  The 3-term pi/2 ends up being used
for about half of its usable range (about 64K values on each side).
2005-10-09 04:29:08 +00:00
..
2005-04-22 18:57:32 +00:00
2005-10-03 14:43:27 +00:00
2005-07-13 10:40:07 +00:00
2005-02-13 23:45:54 +00:00
2005-10-04 22:00:35 +00:00
2005-05-24 10:10:38 +00:00
2004-12-29 02:18:24 +00:00
2005-10-03 14:21:14 +00:00
2005-09-14 10:42:51 +00:00
2004-12-21 10:49:29 +00:00
2005-06-04 10:48:21 +00:00
2004-12-21 10:16:04 +00:00
2005-10-04 08:28:46 +00:00
2005-09-26 06:23:43 +00:00
2005-09-19 08:13:43 +00:00