A Pipelined Processor

12/3/98


Click here to start


Table of Contents

A Pipelined Processor

Outline of Today’s Lecture

Pipelining: Its Natural!

Sequential Laundry

Pipelined Laundry: Start work ASAP

Pipelining Lessons

Overview of a Multiple Cycle Implementation

Multiple Cycle Processor

The Five Stages of Load

Key Ideas Behind Instruction Execution Pipelining

Pipelining the Load Instruction

The Four Stages of R-type

Pipelining the R-type and Load Instruction

Important Observation

Solution: Delay R-type’s Write by One Cycle

The Four Stages of Store

The Four Stages of Beq

A Pipelined Datapath

The Instruction Fetch Stage

A Detail View of the Instruction Fetch Unit

The Decode / Register Fetch Stage

Load’s Address Calculation Stage

Load’s Memory Access Stage

Load’s Write Back Stage

A More Extensive Pipelining Example

Data Hazards

Data Hazard Solution: Register Forwarding

The Delay Branch Phenomenon

Reducing Branch delays (cont.)

The Delay Load Phenomenon

Delayed Load and Branch on a Real MIPS Processor

CPU design Summary

Single Cycle, Multiple Cycle, vs. Pipeline

Additional Notes

Email: alvy@cs.duke.edu

Home Page: http://www.cs.duke.edu/~alvy