Rev 4.12

This is the special release which fixes the subtraction bug with very small
numbers that causes the result to go to negative infinity.  This fix affects
the timing of many operations resulting in about a 5% time increase.  This 
might break some applications, but since it is critical to have correct math
operations this will be released to production and will probably start shipping
in late March or April 01.  If customer's complain about longer scan trigger 
times, we will backdate them to 4.09 flash  

The smallest number before the math underflows to zero is now 1.175E-38.  Note
that operations like A = A - 0.7*A will never evaluate to exactly zero since 
0.7*A will underflow to zero leaving A unchanged.

03-22-01 

