Machine cycles in x86 CPU after assembly instruction

Question

Theoretically I have Intel 8086 and I want to run this instruction:

add [2000], 6 (or in AT&T syntax: addw $6, 2000)

I also know that:

• at 2000 value is 2
• at 3000 is register CS
• at 6000 is register DS
• at 1000 is register IP

Now I want to describe machine cycle in this CPU: fetch & execute.

I know how it looks in theory:

Fetch:

• fetch an instruction from memory
• decode the instruction to determine the operations
• fetch data from memory if necessary

Execute:

• perform the operation on the data
• store the result in memory if needed

But I don't know how it looks with real CPU registers and assembly code. Can you explain this?

Answer 1

The values in memory or registers don't matter. When

        add     word ptr ds:[02000h],00006h

is executed, the cpu reads 16 bit value 2 from location 2000 into an internal (unnamed) register, adds 6 to the 2, and stores and stores 16 bit value 8 into location 2000 (little endian format). At least two web sites shows this as taking 23 cycles (17 for instruction + 6 for the effective addressing).