Simplify pow(2.0, itofp(y)) to ldexp(1.0, y).
Does this require fast math due to rounding in the sitofp conversion? Off the top of my head, probably it doesn't matter because those cases would produce zero or infinity anyway, but it's worth explaining in the code.
|1327 ↗||(On Diff #214395)|
Can we extract this “fits in int” code to helper function? optimize exp2 checks it too, and that fold could use it too... and my powi fold too