Hi, as far as I understand when I multiply two 32 bit integer numbers, the processor produces a 64 bit result, but then the compiler truncates the upper 32 bits to produce a 32 bit value. Is there a way in gcc to avoid this truncating? The problem is that I need to multiply two potentially big 32bit numbers(I don't want to use long long because it's slow) and need to verify whether their product fits into 32 bit integer.
Thanks !
I don't know about a solution offered by GCC, but you could write a function for such multiplication. Let it receive two 32 bit variables, design a 64 bit datastructure and store the result of the multiplication in that datastructure.
Well, my performance grows by at least 20% if I use ints instead of (long long)
Right now I'm doing the following:
long long result = (long long)a * (long long) b;
if (result > MAX_INT) then overflow
else no overflow.
but that's still not fast enough, considering that the CPU can do it all for me..
