They seem exactly the same to me, but the wikipedia page on Superscalar CPU architectures says
While a superscalar CPU is typically also pipelined, pipelining and superscalar architecture are considered different performance enhancement techniques.
Asked
Active
Viewed 1.7k times
3
phuclv
- 30,396
- 15
- 136
- 260
1 Answers
6
The Wikipedia articles for pipeline and superscalar explain these quite well.
Definition before we start: Functional Unit: Hardware device that performs a specific instruction
- E.g. Adder for
add/sub, Multiplier formul, Shifter forsll/slr, Divider fordiv, etc.
Simply put, a pipeline starts the execution of the next instruction before the first has completed - but all instructions are still executed in series and in order.
- Series = Not parallel = not "all instructions start at the same time"
- Only one FU per operation-type is needed. Only need 1 adder and 1 multiplier and 1 shifter, etc.
A superscalar architecture can start two or more instructions in parallel in a single core (using multiple, repeated Functional Units that each handle a specific type of instruction), and independent instructions may get executed out-of-order.
- Having three Adder FUs means you can execute three different
addinstructions simultaneously, but you can't perform any extra multiplications, shifts, etc because you're missing the necessary hardware.
Extra Note: If you're aware of Tomasulo's algorithm, note that the Issue stage is different than the Execution (and WriteBack) stage; above mentions specifically the Execution stage, not the entirety of Issue+Execution+WriteBack stages.
tl;dr
- Pipeline = Serial on single-core (but more instructions pass through than 1-instruction-at-a-time)
- Superscalar = Parallel on single-core
