The MASM Forum

General => The Campus => Topic started by: DanWebb314 on June 09, 2012, 03:32:37 AM

Title: unsigned add
Post by: DanWebb314 on June 09, 2012, 03:32:37 AM
How can I add two unsigned integers?  'ADD' uses signed addition.  Is there an instruction  to do an unsigned add?
Title: Re: unsigned add
Post by: dedndave on June 09, 2012, 03:34:22 AM
same instruction
that is the nature of two's compliment

be aware that if you add 2 unsigned values, and the result is larger than the register, it will set the carry flag
if the result fits in the register, the carry flag will be cleared
Title: Re: unsigned add
Post by: zooba on June 09, 2012, 10:15:17 AM
As dedndave says, the same instruction is used. The difference is in the flags that should be used to interpret the result:

When an unsigned addition overflows (MSB goes from 1 to 0) or an unsigned subtraction underflows (MSB goes from 0 to 1), the carry flag (CF) is set.
The JC/JNC instructions let you branch based on this, as do JA/JAE/JB/JBE ("above" or "below" when you've used SUB or CMP).

When a signed addition overflows (MSB goes from 0 to 1) or a signed subtraction underfloms (MSB goes from 1 to 0) the overflow flag (OF) is set.
The JO/JNO instructions let you branch based on this, as do JG/JGE/JL/JLE ("greater" or "less" when you've used SUB or CMP).

Two's complement enables an identical representation to be used for signed and unsigned integers, and identical electronics to implement addition and subtraction (but not multiplication and division).

Cheers,
Zooba  :t
Title: Re: unsigned add
Post by: DanWebb314 on June 13, 2012, 07:34:54 AM
Very helpful
Thank you
Title: Re: unsigned add
Post by: KeepingRealBusy on June 13, 2012, 08:19:39 AM
DanWebb314,

Note that you can use the same two instructions (add and sub along with two others) to do 64 bit arithmetic:

Code: [Select]
.data
    FirstLow    dd 12345678h
    FirstHigh    dd 87654321h
    SecindLow  dd 01111111h
    SecindHigh dd 00111111h
.code
    mov eax,FirstLow
    mov ebx,FirstHigh
    mov ecx,SecondLow
    mov edx,SecondHigh
;
;   to add
;
    add eax,ecx    ;   Causes a possible carry
    adc ebx,edx    ;   Adds the two numbers plus the possible carry.
;
;   to subtract
;
    sub eax,ecx    ;    Causes a possible borrow
    sbb ebx,edx    ;    subtracts the two numbers and a possible borrow.
;
;   eax ebx should be same as FirstLow and FirstHigh
;

Dave.
Title: Re: unsigned add
Post by: DanWebb314 on June 13, 2012, 08:55:28 AM
Cool!
Title: Re: unsigned add
Post by: KeepingRealBusy on June 13, 2012, 09:15:20 AM
You can increment a 64 bit number by:

Code: [Select]
    add    FirstLow,1
    awc    FirstHigh,0

Dave.
Title: Re: unsigned add
Post by: jj2007 on June 13, 2012, 04:11:00 PM
Quote from: KeepingRealBusy on June 13, 2012, 09:15:20 AM
    awc    FirstHigh,0

Typo, it should be adc aka add with carry. Note it won't work with inc FirstLow because inc does not set the carry flag.
Title: Re: unsigned add
Post by: KeepingRealBusy on June 14, 2012, 12:48:29 AM
Jochen,

Thank you for catching that, don't want to propagate bad code.

Dave.
Title: Re: unsigned add
Post by: jj2007 on June 14, 2012, 04:10:28 AM
Quote from: KeepingRealBusy on June 14, 2012, 12:48:29 AM
...don't want to propagate bad code.

Don't worry, ML.exe would have stopped you with a nasty error message :biggrin:
But the inc thing is worth remembering - I think the other Dave pushed me onto that one a long time ago :P