Einherjer88
Junior Member
- Apr 23, 2008
- 1
- 0
- 0
Hi!
But wouldn't it be more logical that there is, if even so, the same or less input lag?
Imagine the GPU is still rendering on one frame, with this setting to lets say 5, the CPU precalculates during this time in best case this 5 frames and they're queed up right ready for GPU to take them. For GPU there would be no change, cause as soon as it finishes rendering the one frame, it takes out the first of the 5 already from CPU precalculatet frames to render this frame.
There would still be 4 more allready calculated frames in que and cpu just needs to calculate one more frame for filling up que to 5 and then has time to play chess against ram or something else...
This cenario could be in a game with not much geometry (work for CPU) but much effects, lightsources,.... (much to render on one frame for GPU when got CPU geometrysetup) where GPU needs in this example 4-5 times longer for finishing one frame than the CPU.
So where would there be more input lag? Because if the GPU needs so long time to calculate one frame it should make no difference if there are waiting 2 or 4 (depending on new driver setting) more frames in CPU precalculated que....
But wouldn't it be more logical that there is, if even so, the same or less input lag?
Imagine the GPU is still rendering on one frame, with this setting to lets say 5, the CPU precalculates during this time in best case this 5 frames and they're queed up right ready for GPU to take them. For GPU there would be no change, cause as soon as it finishes rendering the one frame, it takes out the first of the 5 already from CPU precalculatet frames to render this frame.
There would still be 4 more allready calculated frames in que and cpu just needs to calculate one more frame for filling up que to 5 and then has time to play chess against ram or something else...
This cenario could be in a game with not much geometry (work for CPU) but much effects, lightsources,.... (much to render on one frame for GPU when got CPU geometrysetup) where GPU needs in this example 4-5 times longer for finishing one frame than the CPU.
So where would there be more input lag? Because if the GPU needs so long time to calculate one frame it should make no difference if there are waiting 2 or 4 (depending on new driver setting) more frames in CPU precalculated que....