A Vector Pipeline for Floating-Point Addition
An example showing the stages of a floating-point addition:
s = x + y, where x, y, and s
are vectors (one-dimensional arrays) containing 32
floating-point values.
This sample vector pipeline shows six stages:
- Stage A:
The exponents of the two floating-point numbers to be added
are compared to find the number with the smallest magnitude.
- Stage B:
The significand of the number with the smaller magnitude is shifted so that
the exponents of the two numbers agree.
- Stage C:
The significands are added.
- Stage D:
The result of the addition is normalized.
- Stage E:
Checks are made to see if any floating-point exceptions occurred
during the addition, such as overflow.
- Stage F:
Rounding occurs.
The number and purpose of stages in vector pipelines vary
for architectures produced by different manufacturers.