When the data to be rendered is stored in RAM on the client side, the rendering is called immediate mode, because the client is asking the server to render the data immediately after it is sent to the server. 

Former examples of immediate mode rendering include

• glBegin... glEnd: pass vertices to GPU forming one primitive at a time
• Vertex arrays:  pass vertices to GPU in a single call (for each frame)

Rendering with glBegin…glEnd is particularly slow because the graphics hardware must wait for each primitive to be sent before it can be rendered!

In contrast, retained mode rendering passes geometry stores the vertices in VRAM on the GPU.  Rendering is often much more efficient. 

Immediate mode rendering was deprecrated in OpenGL 3.0 and dropped from the core in OpenGL 3.1.

High