Blur Busters Forums

How much input lag does msaa, ssaa etc actually add to a game? Is it also dependent on the multiplier?

In know that fxaa doesnt have a real impact but what about the others?

basically depends on how long it takes to render a frame

for instance if aa drops your fps from 1000 to 500 then you get 1ms more input lag

Youre sure? What if i have a frame cap like in csgo of 300 and it stays the same with or without AA. What happens then?

treach wrote:Youre sure? What if i have a frame cap like in csgo of 300 and it stays the same with or without AA. What happens then?

In theory, input lag would be the same. In practice, you can still get more input lag compared to AA off, since AA needs more time to render, so with a framerate cap, the rendered frames will be based on somewhat older input.

Usually this is totally unnoticeable. However, bugs or bad design in the game can actually lead to more input lag than there should be if you crank up the graphics details.

This is correct.

GPU share of input lag is proportional to frametime -- the amount of time it takes to render a frame. Enabling AA increases frametime, which reduces framerate.

Thus, the lower the framerate (during VSYNC OFF) on the same GPU, the more input lag you will get.

However, at 300fps, the input lag share is pretty tiny (300fps = 1/300th of a second per frame) and if you can maintain 300fps with AA, then you can keep it on if you really like it. If you slow down from 500fps (2ms per frame) to 300fps (3ms per frame) during AA, then you'd be adding only 1ms input lag from enabling AA. Also, there are also many settings that will increase framerate, like reducing shadow detail, less shader effects, etc.

Capping framerate versus uncapping framerate, doesn't affect frametime.

treach wrote:Youre sure? What if i have a frame cap like in csgo of 300 and it stays the same with or without AA. What happens then?

if the game's framerate cap works like this (and I sure hope it does): then the input lag will be related to how long it takes to draw the frame. so if it takes 1ms to draw without AA and 2ms to draw with AA, then AA will add 1ms of input lag, even if fps is capped to 300 or whatever

Thats what i mean flood, it cant be only dependent on frametime because the picture is processed, the actual time it takes is unknown?

you can estimate by running it completely uncapped where there is no sleep involved.

there must be more to it. the mouse for instance becomes considerably more sluggish with aa on. nearly in every game. the framedrop alone cant explain that.

paulemannsen wrote:there must be more to it. the mouse for instance becomes considerably more sluggish with aa on. nearly in every game. the framedrop alone cant explain that.

Single-GPU setup or SLI setup? SLI setups have some additional complex considerations relating to input lag, as frame rendertime can actually exceed (1/fps), which complicates the input lag math a little.

Also, if you're running a framerate cap while having AA on, then this can definitely happen as frame rendertime goes dramatically up even if the framerate is the same.

Also, if your framerate is usually consistent but drops more frequently to unplayable levels during busy moments, it will obviously affect lag. e.g. 20fps versus 30fps during slowdown -- can have a dramatic difference in input lag feel, even if normal framerate is usually consistently high.

Also, certain types of AA, that uses a high resolution framebuffer and downsampling, you may have scaling-related input lag. So that might be what you are seeing. Certain forms of AA may add input lag via that step.