32 bit computers have 32 bit registers. If you were to add two 32 bit integers together and the result takes more than 32 bits of space the remainder goes into another register. I don't remember which register that is however.
EDIT:
It's the FLAGS register.
Last edited by ITAmember; 06-29-2009 at 11:42 AM.
Well, when you do an add there's only one bit that can carry, so that just sets the carry flag (it doesn't set the entire register). Multiplication takes two registers (the usual accumulator eax gets the least significant part, I think, and then one of the other registers gets the high part).Originally Posted by ITAmember
32 bit computers have 32 bit registers. If you were to add two 32 bit integers together and the result takes more than 32 bits of space the remainder goes into another register. I don't remember which register that is however.
EDIT:
It's the FLAGS register.
Yep in the registers.
Because two byte-sized values can add up to give a number that's greater than a byte in length. So the micro fills up the register with sign bits(or not) if unsigned hasn't been specified when declaring the type of the variable.
Edit: forgot to say that this happens for chars only because of default promotion rules built into the compiler.
Last edited by itCbitC; 06-29-2009 at 02:58 PM.
EDX I believe.one of the other registers gets the high part).
When you assign a char to a 16-bits short for example, the compiler will generate instructions to do sign extension or zero extension (depending on whether char is by default signed or not). For example, the MOVSX (move with sign extension) and MOVZX (move with zero extension) instructions. And of course, these instructions take clock cycles.I was wondering .. where do these conversions take place ? im sure they won't be at the memory level because that will cause overriding and overflows in memory ... is it in the registers ?
C is a high level language (compared to assembly), and there are hidden costs everywhere.
They are converted when you tell the compiler to (when you assign a char to an int for example).Anyway why should they be converted afterall?Whats the purpose behind it.
