I need the algorithm or program that tells how much to shift and what to add after the shift when dividing or multiplying by a certain number.

For example
mov eax, dword[SomeNumber]
mov ecx, ????
div ecx
/\/
mov eax,dword[SomeNumber]
shr eax, ????
add eax, ????

Thanks for the help in advance

Search for AMD's optimization manual.

UPDATE: The file I have is named 22007.pdf, but I guess there's a newer version of it with a different name. Title: "AMD Athlonâ„¢ Processor - x86 Code Optimization Guide". I could post the (small) code here, but you'll appreciate read the manual.

howdy r22
its a problem , because you cannot always replace a mul by shl and add in the way you wanna.

you can use this optimization only if you are multiplying by the same number, ex.: 10, or 320

The Svin's magic divider will convert div to mul and shifts. Source code included.

You should check that site I have as a signature.
It is NOT in vain my idol
direct link to MUL:
http://www.azillionmonkeys.com/qed/amult.html
---and DIV:
http://www.azillionmonkeys.com/qed/adiv.html

P.S. you might have interest in SQRT too:
http://www.azillionmonkeys.com/qed/sqroot.html