CSE 30 -- Lecture 3 -- Oct 7

• Homework 1 clarifications:
  • it is okay to either specify non-instruction memory contents via notation like mem[5]=1 or via MIPS assembler-like directives. See lecture 2 notes for an example. You can assume that data placed in memory via a .data directive or instructions placed in memory via a .text directive will not coincide with the input values x and y in memory locations 0 and 1 respectively, nor with memory address 2 (this is where the output goes). You may use symbolic names for these instead too, as long as you comment your code to this effect.
  • efficiency for this assignment does not matter -- how fast your code runs will matter in future assignments, but not for this one. If you look carefully, the example in the lecture 2 notes has a couple of instructions that can be omitted. The goal of the One-Instruction Computer is to learn how to make longer, more complex code sequences out of simple instructions, learn basic assembler syntax, talk about macro assemblers, and learn that there is a universal instruction set (much like NAND is universal for circuits), not to actually use it as a real instruction set.
• One-instruction computer: how to do subroutine call/return with self-modifying code.
• Why self-modifying code (on a real machine) is a bad idea in real computers.
• Isn't the OIC the RISC idea taken to an extreme?

bsy@cse.ucsd.edu, last updated Wed Oct 16 22:37:37 PDT 1996.