Hi Ryan,
Some older devices don’t support full precision floats at all. For those that do, full floats are generally half the speed of half floats, but we realize there are cases where you want to eat the cost, and we have a feature request currently slated for 4.13 that will enable you to turn on full precision on a per material basis. The ticket for that is UEMOB-109.
In the interim, you could try use the “CustomUV” feature of the material system which lets you perform calculations in the vertex shader and send the results to the pixel shader, which does allow you to work around a lot of precision issues.