The addressing formula: [base + scale·index + disp]

Every iteration m0 reads sequentially from the pointer, while m1 aims via the full formula. Hit "Step" and watch the address assemble itself.

[base+scale·index+disp]
base — a pointer register (here: srcq) scale — only 1, 2, 4 or 8 · index — a register (here: r1q) disp — a constant known at assembly time (here: 3+mmsize) 

    

    

      
      
      
      
    

  


  

    
Address calculator — where m1 reads from

    

      [
      
base · srcq—

      +
      
2 · r1q—

      +
      
disp · 3+1619

      ]
      =
      
address—

    

    
disp 3 + mmsize is arithmetic between constants: the assembler folded it into 19 at build time. The CPU receives a ready-made number.

  


  

    
Memory (RAM) — bytes at addresses 1000–1079

    

  


  

    
CPU — registers

    

      

        
srcq
base — "where from"1000

        
r1q
index and counter?

        
after dec: —

      

      

        
m0empty

        
m1empty

      

    

  


  

    m0 window — 16 bytes from srcq
    m1 window — 16 bytes from the computed address
    srcq — the pointer
    dimmed — already passed by the pointer