exp:
exp is the ratio of the number of floating point exponential and logarithmic (i.e. expf & logf) instructions to the total number of floating point instructions.

Classifying exp:
exp can be classified into low, medium or high as follows:

Bucket	Condition
Low	The region of code contains no exp instructions. OR 1 or fewer instructions out of every 4 floating point instructions is a exp instruction.
Medium	1 out of every 2 floating point instructions is a exp instruction.
High	2 or more instructions out of every 3 floating point instructions is a exp instruction.

Example:
Consider the following code:

for (i=0; i<1000; i++)
   result[i] = expf(arrayOne[i]) * arrayTwo[i];