/*When I was a kid with a Commodore 64,
the 6502 chip had no multiply inst
and this is how we had to do it, except,
I used more regs in this example.
*/
asm {
//Opcodes are slightly different to make writing my x86_64 assembler easier.
//See ::/Compiler/OpCodes.DD.

//You can clobber RAX,RBX,RCX,RDX,R8,R9.  The compiler expects that.

MUL_BY_HAND_U8_U8_TO_U16: //This is only for fun.
//8bit * 8bit-->16bit
//AL*BL-->AX
        MOV     CL,8
        SHL     AX,8
@@05:   SHL1    AX
        JNC     @@10
        ADD     AL,BL
@@10:   DEC     CL
        JNZ     @@05
        RET

_MUL_BY_HAND_U8_U8_TO_U16::     //C callable
        PUSH    RBP
        MOV     RBP,RSP
        MOV     AL,U8 SF_ARG1[RBP] //SF_ARG1
        MOV     BL,U8 SF_ARG2[RBP]
        CALL    MUL_BY_HAND_U8_U8_TO_U16
        MOVZX   RAX,AX
        POP     RBP
        RET1    16

_MUL_U64_U64_TO_U128::
//64bit * 64bit-->128bit
        PUSH    RBP
        MOV     RBP,RSP
        MOV     RBX,U64 SF_ARG3[RBP]
        MOV     RAX,U64 SF_ARG1[RBP] //SF_ARG1
        MUL     U64 SF_ARG2[RBP]        //Res RDX:RAX 128bit
        MOV     U64 [RBX],RAX
        MOV     U64 8[RBX],RDX
        POP     RBP
        RET1    24
};

//My convention is to put an underscore
//on C callable asm routines.
_extern _MUL_BY_HAND_U8_U8_TO_U16 U16 MulU8(U8 n1,U8 n2);

class U128
{
  U64 lo,hi;
};

_extern _MUL_U64_U64_TO_U128 U0 MulU64(I64 n1,I64 n2,U128 *_prod);

U0 MulByHand()
{
  U128 p;
  "2*7   =0x%X\n",MulU8(2,7);
  "100*10=0x%X\n",MulU8(100,10);

  MulU64(0x0123456789ABCDEF,0x1000001,&p);
  "0x0123466789ABCDEF*0x1000001=0x%016X%016X\n",p.hi,p.lo;
}

MulByHand;