opengl - What is tile based deferred rendering and what is its advantage? -

i read concept here.

but did not part: "the key advantage deferred renderer accesses memory efficiently. partitioning rendering tiles allows gpu more cache pixel values framebuffer, making depth testing , blending more efficient."

is tile based rendering happens in parallel? rendering multiple tiles @ time different cores?

also advantage of normal tile-based rendering?

the mobile gpu apple discuss on page contain small amount of memory physically on gpu chip , fast access. splitting render target tiles small enough fit in memory, , processing 1 @ time, minimise amount of interaction slower main memory - rather having fetch, test, blend etc depth buffer , colour buffer values each pixel in each triangle rasterise triangles, rasterise tile fast memory , write each tile's final raster out main memory done it.

additionally, tile based deferred renderer don't rasterise triangles until have calculated triangles visible each pixel/quad in tile, end shading pixels contribute final scene.

there other potential benefits tile based rendering culling, visibility testing , locality of memory accesses on platforms rasterisation not deferred and/or less constrained main memory performance. this article bit of tangent asking, may give better idea of these general benefits while describing in depth how modern rasterizer works.