In Simulink, which option is more efficient, regarding computational cost?

I've got a Simulink model with a division of two signals that are always greater than or equal to 0. I know that this division is 0 if the denominator is 0.

Which option should I choose to obtain the better performance or lower computational cost?

OPTION 1 uses, in my point of view, 1 variable less.

But, on the other hand, OPTION 2 uses a constant signal, instead of the variating in2.

What do you think/suggest?