his calculator follows classical approach when uncertainty of f(x) calculation is estimated by formula max|(derivative(f))|*|x*uncertainty(x)|, where maximum of function derivative is considered on interval [x-uncertainty(x), x+uncertainty(x)]. Thus sin(2*pi)=0+-1E-36 and sin(2*1E20*pi)=0+-1E-16. As we see, the results accuracy degrades with growth of argument, but such approach allows to preserve all trigonometry facts like sin(even number*pi+x)=sin(x). Calculators with multi-precision allows to calculate sin of big argument, like 1E40, with any precision, but cannot calculate sin(1E40*pi) since they dont have pi. The pi becomes for them a floating number with arbitrary precision. It seems strange, because 2*pi corresponds to one rotation and counting rotations is much easier then measuring 1E40 radians.
In the default mixed mode the calculator treats numbers as integers wherever it is possible. For example, 38! have an integer result with 45 digits.
There are 10 variables (or constants) available for storing frequently used numbers.