Forward+
Tiled forward rendering: split the screen into tiles, cull the light list per tile, then shade in a single forward pass that touches only the lights the tile is actually reached by.
Classic forward rendering grows expensive as lights multiply, because every object is in effect tested against every light. Forward+ adds a depth pre-pass and a compute step that, for each screen tile, builds a short list of the lights that reach it. The shading pass iterates only that list, and the cost is decoupled from the scene's total light count.
The method (Harada et al., AMD, 2012) keeps what deferred shading is forced to give up: working MSAA and transparency, plus lower memory bandwidth since no G-buffer is needed. Clustered shading is the natural next step and fixes Forward+'s weakness with lights at different depths inside one tile.