Microprogramming

• Current state and outputs are defined using microinstructions
• The microinstructions are stored in microprogram memory (control memory)
• A micro-sequencer computes the address of the next microinstruction using the inputs and the current state

Implementation

• Option: CM stores the complete state transition table
  • A row in the table corresponds to a microinstruction (i.e. next state and outputs)
  • The micro-sequencer selects the next state based on the input one of the next states as address for the CM (uses a multiplexer)
  • 
  • Advantage: Most flexible ($\delta$ and $\mu$ are reprogrammable)
  • Drawback: CM can be very large and slow
• Option 2: Store the instructions by their address and their conditional branch address
  • Moves to next by default, moves to BADR if the selected condition is met
  • 

Generating output signals

• Horizontal microprogramming
  • 1 bit is used for each output signal (a control field) in CMDR → Horizontal microprogramming
  • Advantage: maximal parallelism in data path
    • All control signals can be set independently
  • Drawback: Wide microinstruction → large control memory
• Vertical microprogramming
  • Only one control field with $\lceil ld(m)\rceil$ is used
  • Values corresponds with a binary coding for the full control fields
  • Advantage: small control memory
  • Drawback: only one signal can be set at a time; Slow; No parallelism
  • 
• Diagonal microprogramming
  • Control signals that can not be simultaneously active are grouped together and binary coded
  • The number of fields defines the level of parallelism
  •