Re: Radians Or Degrees?

Michael S wrote:

To make it less tempting, you could try to push for inclusion of
rsqrt() into basic set. Long overdue, IMHO.

Many Newton-Raphson iterations converge more rapidly with RSQRT than
with SQRT, for reasons I never looked that deeply into.

Right now, I can't think of any other transcendental that I really want
to elevate to higher status. It seems to me that elevation of log2(x)
and of 2**x will do no harm, but I am not sure about usefulness.

Ln2 and EXP2 are the basis of My 66000 log and EXP families, performed in
HW function unit.
Ln2 has the property that a binade [1..2) maps directly to another binade
[2..4); exp2 has a similar property. Ln2 is easier to get good accuracy
from than Ln or Log as a starting point. But in any event:: you are always within a high precision multiply of the correct result = round(53×106);