Performance on arrays

[chameleon]

It is a little bit machine specific question.

Which is better approach?

for (int x = 0; x < width; x++)
  matrix[x+width*y]=blabla;

or

for (int x = 0, offs=y*width; x < width; x++)
  matrix[x+offs]=blabla;

I believe the second, but if I remember well, both lines translated to only one x86 assembly instruction.

Am I correct?
Have both instructions the same clock ticks?

If it is true, we use first approach because it is more watchable.

[kev82]

I think that an optimizing compiler will realise that width*y is constant in the loop and optimize out the multiplication. Try it, generate the assembly and see.

I wouldn't worry about it though, if that's the slowest point of your program you're doing very well.

[zb0S216C]

I personally would go for the 2^nd option since the CPU doesn't have to re-compute y * width with every pass of the loop.

_Wazzak