Assembly (NASM) How to do Numerical operations

Question

I programmed a lot in java and know the basics of python and sometimes I play with c++... And those are all High Level Programming Languages and they're great and all but I was curious about how things worked at a deeper level in this case assembly language... I started to learn assembly for the x86 processor and chose NASM as my assembler. I spent some time learning how registers and the stack work, how the information is stored and how the information can be changed , not just reading but practicing, using a debugger, etc and something really started to bother me: NUMERICAL OPERATIONS...

What I mean is whenever something is read from input it is stored as the correspondent ASCII value on the memory and I'm ok with it... I understand the basics of how to read and write to console with the whole pass offset and length system however as I mentioned before the characters including numbers are stored as ASCII values.. 0 is stored as 0x30, 1 as 0x31, etc... That means a single digit is stored as a whole byte and I see that as a disadvantage because I know a byte can represent numbers up to 255 which would save a lot of memory and time to do math operations and leave space for other data... In languages like c++ an integer value is represented by only 4 bytes which means those languages can store integers up to 4294967295

Getting to the point... What I really want to know is what is the best approach to represent numbers and do basic arithmetic operations using nasm.. should I leave those numbers as ASCII values and do the operations on a byte level or should I convert those numbers so a byte can hold a larger value ? Or is there a way to read numbers more efficiently ?

PS.: Sorry for the long post, I just wanted to give an overall background of my concern so you guys can help me based on it... And for the bad english ... I tried to be as clear as I can (English is not my native language but I try to understand it as much as I can... It helps a lot knowing english when I'm programming)

Answer 1

I would transform from ASCII. It makes everything simpler, imagine multiplying ascii numbers. And if you have more than one digit using ascii could become really annoying. Also, you can declare variables of more than one byte:

b=byte

w=word (2 bytes)

dw=double word (4 bytes)

var1 w   0000h ;
var2 dw   0007h
var3 b 0000h

This way you can get input from your console. Just for one digit:

mov ah, 1  ;int 21h (interrupt 21h) in the next line is a function that can do many things
int 21h    ;depending on the value you put in register "ah". Using 1 it reads input the value is stored in "al"
add al, -48 ;ascii -> bin .. here you substract 48 to transform from ASCII
mov [var1],al

You can add two numbers like this (you need something extra for negative numbers) you'll get

[var1]=[var1]+[var2]

mov al,[var2]
add al, [var1]
mov [var1],al

CPUs generally also have instructions for multiplying

check this out, its the instruction set for the 8086 processor, this are all the things that you can tell it to do. 8086 Instruction set

Answer 2

First off, you cannot... I repeat, you cannot compare Assembly to any High Level Language! How they store and interact with data is totally different. They do it their way to make the programmers life easier. If you want to store the ASCII code of a char as a DWORD, go ahead.

In Assembly, there is absolutely no hand holding. The CPU has no idea what you want to do with any data, be it a byte in a file or a character entered on the keyboard.

You do everything yourself in Assembly. The characters you press on a keyboard are ASCII because that is what gets displayed when you type them, if you want to do something else with them, you the programmer converts it however you need.

Converting to an ASCII number to a number is trivial.

Numbers: 0  1  2  3  4  5  6  7  8  9
ASCII:   48 49 50 51 52 53 54 55 56 57

See the difference between them? 48 is the magic number.

You loop through your string, testing each digit to see if is between 48 and 57, anything else skip. If the ASCII char is between 48 and 57, subtract 48 from the character and you have a number. You add 48 to get the ASCII char of the number.

Not telling you to use a atodw/atoi or whatever, you learn more by doing yourself, and it isn't hard.

Answer 3

As in any language: If you have a string but you want to do numerical operations on it, convert it to numeric. Unless it doesn't fit in any general register of your processor.

I don't now what instruction set you are targeting, but eg. in x86-64 there are 8-bit, 16-bit, 32-bit and 64-bit general registers to hold your numbers. If your numbers don't fit it 64 bits (in x86-64) then things can get a little bit more complicated. If you want the biggest range of numbers available in the architecture you are targeting, the best way would be to store the numbers in the biggest possible register or in memory.

In contrast to high-level languages, in assembly level there is no difference between signed and unsigned variables, it only depends on how you handle it. There are signed and unsigned operations though, such as signed multiplication (imul), signed division (idiv), unsigned multiplication (mul), unsigned division (div).

Answer 4

Ascii characters are NOT numbers.

You need to use atoi function.