"Multithreaded game engines also usually have more input to screen lag than single threaded game engines. To get max power out of the current console platforms, multithreading is required.
Usually you are processing game logic and physics on one (or several threads) at the same time as you are rendering (separate render & render setup & culling thread). The rendering must be done using last frame game state, as the current frame game state is not ready yet (physics and game logic threads are processing it on different threads).
Single threaded game game loop:
1. Process input
2. Process game logic & physics
3. Render frame
Multithreaded game loop:
1. Start rendering graphics by using last frame finished game state in it's own thread(s)
2. Process input
3. Process game logic & physics (usually in multiple threads)
(4. Graphics thread is finished around here and updates the visible frame independently on game logic & physics)
5. Synchnonize threads
With the multithreaded setup you get slightly longer input lag. The maximum input lag equals the time it takes the graphics chip to render the frame."
http://forum.beyond3d.com/showpost.php?p=1270913&postcount=27
http://www.gamasutra.com/view/feature/3725/measuring_responsiveness_in_video_.php?page=3
Game lag article
I believe that you're correct, and that this may be one of the under appreciated benefits of the Xbox 360's built-in scaler. I've measured lag using the Rock Band 2 calibration system, and I found the following:
Xbox set to output 1080p directly to my 1080p Aquos LCD TV: 0 ms video lag (very repeatable)
Xbox set to output 1080p to my Yamaha AVR, then the TV: 20 ms lag
It appears that the TV buffers a frame or even two while scaling the picture from 720p to 1080p native resolution, while the receiver fairly efficiently strips the audio from the HDMI and passes the video through to the TV.
I'd like to conduct similar measurements on my PS3, but I have no software with which to do so.
http://forum.beyond3d.com/showpost.php?p=1270437&postcount=5
Tease.
