In my design, the "modified" single-port distributed RAM (RAM with dual-port addressing, but single-port output) uses two LUTs. If I am not using the outputs for a second LUT (which will end with an SP), why does this not get trimmed away?
If the SPO (single-port out) is unused, it is still necessary to use both LUTs. This occurs because even though the SPO LUT is not needed, it cannot be used for anything else as its address lines are bound to the DPO half.
If you want a true single-port distributed RAM, use the Language template and it will use one LUT, and dual-port will use two LUTs.