In terms of the actual execution of operations, the architecture provides four
execution ports (each with its own pipeline) that are responsible for the actual
execution of instructions. Each unit has different capabilities, as shown in
Figure 2.4.
Figure 2.4 Issue ports and individual execution units in Intel NetBurst processors.
Po rt 0
Double Speed ALU
ADD/SUB
Logic Operations
Branches
Store Data Operations
Floating Point Move
Floating Point Moves
Floating Point Stores
Floating Point Exchange (FXCH)
Po rt 1
Double Speed ALU
ADD/SUB
Floating Point Execute
Floating Point Addition
Floating Point Multiplication
Floating Point Division
Other Floating Point
Operations
MMX Operations
Integer Unit
Shift and Rotate
Operations
Po rt 2
Memory Loads
All Memory Reads
Po rt 3
Memory Writes
Address Store Operations
(this component writes the
address to be written into
the bus, and does not send
the actual data).
66 Chapter 2