CS 3220 Processor Design
Fall 2002

Instructor: Pete Manolios

Project 3, Due 12/5/2002 (before midnight)

This is a team project. Each team must consist of 2-4 people. Single person submissions will get an automatic 0.

The point of this project is to implement the IJVM2 in hardware.

The specification of the IJVM2 is ACL2 specific, but since it tells you what needs to be done at the bit level, it is not hard to see that it can also serve as a specification that can be used for an Altera design. There are two issues, however.

  1. In the ACL2 specification, memories are alists mapping addresses to values. In the Altera design, you can implement RAM with Embedded Array Blocks. Look at the documentation on RAM, Embedded Array Blocks, and ROM/RAM Megafunctions. It may also help you to look at chapter 13 of the Hamblen and Furman book.
  2. In order to be able to probe the memory in the Altera implementation, we will add some instructions to the IJVM2. The instructions are used to show the contents of bytes on the LED display of the Altera board. The format of the instructions is as follows.

    Instruction Opcode
    ISHOW1 1
    ISHOW2 2
    ISHOW3 3
    ISHOW4 4

    The instructions work as follows. They use the LED to display the ith byte (i is based on the opcode) of [L[C[pc]]], where the notation I am using is the same as that used to define the IJVM2 and the high order byte is byte 1. Also, the pc is incremented by one after the opcode is seen (so that pc in the above expression points to the byte after the opcode), and the pc is incremented after the instruction is processed (so that it points to the next instruction in memory), thus, until now, this is similar to what a load instruction does. But the ISHOW instructions are very different in the that the next instruction is not automatically processed. Instead, the instructions display the byte on the LED (in hex format), and the machine pauses, with the LED still displaying the ith byte of [L[C[pc]]], until the left pushbutton is pressed on the board (don't forget to account for switch contact bounce in your design). Once it is pressed, the machine resumes executing the program.

Here is what you hand in

  1. A README file with your names and a clear explanation of what files are included, how to check them, what you did and did not do, etc.
  2. An archive of your Altera implementation. You can create an archive by clicking on file, then project, then archive in the MAX+plus II tool. This will create a directory with all the necessary files.
  3. Write two non-trivial programs (e.g., use recursion and a reasonable subset of the instruction set) to compute some well-known functions and test your implementation. Provide a detailed explanation on how we can test your system on the examples you provide. Make this as "push-button" as possible, e.g. , provide a memory that is already initialized with the test programs, so that all we do load your program on the Altera board and look at the LED after it finishes.
  4. Document your design decisions and give me a 1-2 page overview of your project.
  5. Return your Altera boards, cables, etc. You have until 4PM on 12/6/2002 to do this.

Notes

It is important that you have a well-documented system that works. Have regular group meetings and make sure that your design is sound and that everyone is resposible for a reasonable subset of the work. If you run into problems, e.g., say that you design does not fit on the board, make reasonable decisions, e.g., limit the size of memories, reduce the datapath, etc.

Read the instructions on turning in assignments carefully.

You do not have to implement the IMUL instruction.

Extra Credit

You can do the extra credit only if you finish the project.

  1. Extend the netlist language described in project 2 so that you can deal with latches, registers, and memories.
  2. Carry out the above project using your new language. Compare the two approaches.

Sumbission Instructions

Create an archive of the files using tar or winzip (if you want to use something else, ask first to make sure we can unarchive it) and send us the archive (as a single attatchement) via email. The email should be sent both to me and to Vernard (vernard@cc).