[Punchinello]

Hello! I'm a beginner of IBM PC assembly language programming and I'm confused with some concepts here so think I need some help.
What are 32-bit instructions and what are 16-bit instructions?
What is operand-size attribute?
And with the MOVSX/MOVZX instrucion, if the destination operand is a 32-bit register and the source operand is a memory location, how does the machine decide whether to read a byte or a word from that location?
Thank you in advance!

[SaswatPadhi]

(1) Do you mean 16 bit and 32 bit instructions or registers ??
16 bit registers are ax, bx, cx, dx and 32 bit registers are eax, ebx, ecx, edx.

(2) Operand size : Any time a memory reference is given as part of an instruction, the size of the memory operand is either implied or must be specified.
Consider this :

mov ax, ds:bx

Here, the operand-size WORD is implied since the AX register is one word in size.

Now, consider this :

inc ds:bx  ;incorrect !!!!! 
 

This is incorrect since compiler will be confused whether the value pointed to by bx is byte or word in size.
Correction :

inc word ds:bx  ;increment word at [bx] 
 

or

inc byte ds:bx  ;increment byte at [bx] 
 

In the correction, word and byte are operand-size attributes.

(3) Partly answered in (2). Look at the first example above :

mov ax, ds:bx

Here compiler assumes that value of [bx] is word because ax has size 1 word. To override, you can write :

mov ax, byte ds:bx

This will move byte value pointed by bx to ax.

Hope I cleared your doubts.

[mayjune]

very well explained saswat..

[xpi0t0s]

A 32-bit instruction is an instruction that takes up 32 bits, i.e. 4 bytes, whereas a 16-bit instruction is one that takes up 2 bytes.
Lots of CPUs have varying length instructions, even the old Z80 did; most were one byte but some were as long as 4 bytes.

[SaswatPadhi]

Quote:

Originally Posted by mayjune

very well explained saswat..

Thanx for the feedback !