Embedded Computer Architectures

 

 

•         Lecturers:

—     Gerard Smit

—     Andrι Kokkeler

 

•         Goal: Students are able to translate requirements for an embedded system into an efficient computer architecture.

—     Requirements:

–        Functionality

–        Reliability

–        Timing

—     Efficient:

–        Power

–        Design and Production costs

–        Design time

 

•         Plenary sessions

—     Session 1: Introduction, CPU Structure and function

—     Session 2: Reduced Instruction Set Computers

—     Session 3: Instruction-level parallelism and superscalar processors

—     Session 4: The IA-64 Architecture

—     Session 5: Parallel processing

—     Session 6: Diverse architectures

—     Session 7: Fine Grain Reconfigurable Architectures

—     Session 8: Coarse Grain Reconfigurable Architectures

—     Session 9: Heterogeneous Architectures

—     Session 10: Associative Processing

 

•         Assignment

—     Select three computer architectures (e.g. Pentium, MIPS, FPGA)

—     Select one application domain (e.g neural networks)

—     Study mapping of the application onto architectures

–        Presentation

–        Paper (no research paper)

 

     

·        Test

o       has to be passed to do the final assignment

o       Final grade just based on final assignment (not on the test)

 

•         Contents paper/presentation

—     Explain basic principles of the 3 architectures

–        Strong points

–        Weak points

–        Scalability

–        Preferred applications

–        Non-preferred applications

—     Give survey of the critical points of the application

—     How well does the application map onto the architectures

 

 

•         Sheets of the lectures

—     Introduction

—     Hennessy and Patterson, Appendix A

—     Hennessy and Patterson, Chapter 1

—     Hennessy and Patterson, Chapter 2

—     Hennessy and Patterson, Chapter 3

—     Hennessy and Patterson, Chapter 4

—     Hennessy and Patterson, Chapter 6

—     Alternative Architectures

–        IRAM

–        Dataflow Machine

–        Raw Processor

–        Streaming Processor

–        Cell Processor

—     Heterogeneous Architectures

—     Networks-on-Chip

—     Associative Processing

—     Image Quantization

—     Data Mining

—     Reconfigurable Computing