saturation arithmetic?

[Cazadorro]

I'm trying to use vectorized saturation arithmetic (ie, if the result of an operation would be larger than the bounds of the type, clamp it to min and max). Opencv offers this out of the box, so I figured I'd see it here, but I can't find it. I can use x functions but I'm afraid of not getting proper SIMD and TBB backend support by doing that.

The operation I'm using this for is extremely memory bound, so I can't afford for this to transform into a simple for loop and not saturate memory throughput. Is there a way to get saturation arithmetic to work with out ruining backend optimizations?