COP 3402 meeting -*- Outline -*-

* Other VM designs

** Designs seen previously

------------------------------------------
        VM DESIGNS PREVIOUSLY SEEN

 1-register machine == Tiny VM


 SRM (3 registers) == VM of Homework 1



------------------------------------------

      ... accumulator is target of all computation
          (limited support for subroutines)

      ... use registers to support computation
          (only JAL helps subroutines)

** Simplified RISC Machine (SRM)
------------------------------------------
    SIMPLIFIED RISC MACHINE (SRM)

32 registers

Num.    Usage             Name(s)
=====================================
  0     always 0          $0
  1     assembler temp.   $at
  2,3   function results  $v0, $v1
  4-7   function args     $a0 - $a3
  8-15  temporaries       $t0 - $t7
  16-23 saved temporaries $s0 - $s7
  24,25 temporaries       $t8, $t9
  26,27 reserved for OS
  28    globals pointer   $gp
  29    stack pointer     $sp
  30    frame pointer     $fp
  31    return address    $ra

------------------------------------------
        Can either use the symbolic name ($t1)
        or the numeric equivalent in assember ($9)

------------------------------------------
         INSTRUCTIONS

All fit in one word (32 bits)
  Register format:
    op:6 rs:5 rt:5, rd:5, shift:5 func:6

  Immediate format:
    op:6 rs:5 rt:5  immed:16

  Jump type:
    op:6   addr:26
    
Register Format
 Arithmetic and Logic (3 registers):
   3 registers: ADD, SUB
                AND, BOR, NOR, XOR



   2 registers: MUL, DIV
   1 register: MFHI, MFLO

 Shifts (2 registers + shift): SLL, SRL

 Jump to register (1 register): JR

Immediate Format
 Arithmetic (2 registers & immediate):
    ADDI


 Logical (2 registers & immediate):
    ANDI, BOI, XORI


 Comparison & jump (2 registers & offset):
   BEQ, BNE

 Comparisons to 0 (1 register & offset):
   BGEZ, BGTZ, BLEZ, BLTZ



Load, Store name 1 register:





Call, Return use static links

------------------------------------------
        ... e.g., SUB s, t, d
                  is GPR[d] <- GPR[s] - GPR[t]

        ... e.g., ADDI s, t, i
                   GPR[t] <- GPR[s] + sgnExt(i)

        ... e.g., BOI s, t, i
                   GPR[t] <- GPR[s] \/ zeroExt(i)

        ... e.g., BEQ s, t, o
                   if GPR[s] == GPR[t] then PC <- PC+formOffset(o)

                   where formOffset(o) = sgnExt(4*o)

        ... e.g., BGEZ s, o
                   if GPR[s] >= 0 then PC <- PC+formOffset(o)

                   so these will jump relative to the PC (== old(PC)+4)