1)
Floating point types have rounding problems. “float”
accuracy to number of digits is almost 7, whereas number of significant digits
for “double” is hovering around 15.
2)
Calculation speeds for integer arithmetic is
much faster (could be double or more) than “float” and “float” faster than “double”
and “double” faster than “decimal”. I mean, much much faster!
3)
So, when should you pick “decimal”. When you want
3.14 be 3.14 and not 3.14000001. “decimal” type doesn’t have rounding problem
like floating point types. But it could cost you. Let’s say, If you have a
billion adds/subtracts, it could slow down your program for more than 100 times
as compared to “double”.
4)
Most computational functions like exponents,
sines are not available for the “decimal” type or at least you have to do an
explicit conversion at some point which means possible “rounding off”.
