Thread: Why does this UL int appear to exceed its limit?

Someone please correct me if I'm wrong, but I see a couple of problems here.

First, rand() and srand() are functions defined in stdlib.h. Are you trying to implement your own versions of these functions? Second problem I see is you are declaring your function prototypes inside of main():

Code:

#include <stdio.h>

int main()
{
       int rand(void);
       void srand(unsigned int);
       ...........
}

They should be defined outside of the main() loop like:

Code:

#include <stdio.h>

int rand(void);
void srand(unsigned int);

int main(void) {
       .......
	
}

Also you should use "void" as the argument to main() if you aren't going to pass any command line arguments, as in the example I provided earlier. Now, onto your original question. The value of next is being assigned the difference of the result of the expression:

Code:

next = next * 1103515245 + 12345; //which is 5517588570

and the size of the data type (unsigned long int):

Code:

5517588570 - 4294967296 = 1222621274

The reason next is being assigned 1222621274 is because of two's complement, if I'm not mistaken.

1222621274 is is then divided by 65536, which gives the answer of 18655. Since you are typecasting the return value as an int, and the result of the modulus division is less than 1, it ignores everything after the whole number 18655.

Hope this helps

Originally Posted by matrixx333

Someone please correct me if I'm wrong, but I see a couple of problems here.

First, rand() and srand() are functions defined in stdlib.h. Are you trying to implement your own versions of these functions? Second problem I see is you are declaring your function prototypes inside of main():

Neither of those is a problem. One is allowed to defined their own version of certain functions and not include the relevant header file that normally gives those function definitions. What is shown here looks like an average rand() implementation from 10-20 (or more) years ago.

Secondly, whilst declaring function prototypes inside a function is not the normal way to do it, there isn't actually a problem with doing it that way in the example posted.

Also you should use "void" as the argument to main() if you aren't going to pass any command line arguments, as in the example I provided earlier.

If this is being compiled as C then yes, but people also frequently post in the wrong forum.

The reason next is being assigned 1222621274 is because of two's complement, if I'm not mistaken.

The actual answer to the question is that it doesn't exceed 4294967295 because the rand() function makes strategic use of arithmetic overflow. Several of the upper bits are lost on each multiply & add.

How is this code processed then, such that the function 'next' doesn't exceed its limit? How are upper bits lost and the calculation still successfully made?

That's just how the CPU functions. It adds bits just like you and I would (except in binary), by adding corresponding digits and keeping track of the carry. When it gets to the most significant bit (so in the case of unsigned int, 2^31) there is no more room for the carry, so it is ignored.

Say you're dealing with four-bit numbers and you want to multiply six by ten.

In binary six = 0110 and ten = 1010.
The result would normally be sixty right? Which is 111100 in binary

But because the variables are only four-bit, all we store is the lower 1100 and the top two ones are simply thrown away, resulting in a decimal value of just twelve.