Page 1 of 1
Jorimt can you test CPU bottleneck with G-Freesync
Posted: 15 Jul 2022, 10:32
by Crazyness
i just typed a novell in another thread,so i am just gonna open this topic and edit post later.
there is not a single topic in online world about cpu bottleneck and input lag
and how its affect rendering pipeline, buffers , and all that jazz.
how can you even test that and how can you measure that..etc
my personal opinion is that,
there will be 0 lag..
completely opposite from Gpu bottleneck
Re: Jorimt can you test CPU bottleneck with G-Freesync
Posted: 15 Jul 2022, 13:25
by jorimt
Both CPU and GPU bottlenecks originate from the system, thus VRR has no effect on either, since it just syncs whatever is output by the system.
As for CPU vs. GPU bottlenecks, a CPU bottleneck, be it cause by limited system and/or game engine performance (and sometimes both simultaneously), will typically result in a lower achievable average framerate and low GPU utilization.
I.E. generally, the only way CPU utilization can directly affect latency is through frametime; the lower the achievable frametime/the higher the average framerate, the less CPU-side latency there will be.
Re: Jorimt can you test CPU bottleneck with G-Freesync
Posted: 06 Sep 2022, 15:49
by Crazyness
jorimt wrote: ↑15 Jul 2022, 13:25
Both CPU and GPU bottlenecks originate from the system, thus VRR has no effect on either, since it just syncs whatever is output by the system.
As for CPU vs. GPU bottlenecks, a CPU bottleneck, be it cause by limited system and/or game engine performance (and sometimes both simultaneously), will typically result in a lower achievable average framerate and low GPU utilization.
I.E. generally, the only way CPU utilization can directly affect latency is through frametime; the lower the achievable frametime/the higher the average framerate, the less CPU-side latency there will be.
as usual jorimt doing his thingy
yes i was aware of that,but that was not the point of my post..i am pretty certain..bcs i alredy knew that..
so maybe i was poorly formating the question..i am not sure.
funny thing..if you for example select in nvcp normal instead max performance under power menagment settings
you can squeeze 20-30 more fps in cpu bound scenarios..
honestly idk why that is a thing,maybe bcs they are more on level playfields cpu=gpu vs cpu-+gpu(higher clocks)
i cant remeber what i originaly had in mind with this post..sure i will remember and post again.
Re: Jorimt can you test CPU bottleneck with G-Freesync
Posted: 06 Sep 2022, 17:59
by jorimt
Crazyness wrote: ↑06 Sep 2022, 15:49
funny thing..if you for example select in nvcp normal instead max performance under power menagment settings
you can squeeze 20-30 more fps in cpu bound scenarios..
honestly idk why that is a thing,maybe bcs they are more on level playfields cpu=gpu vs cpu-+gpu(higher clocks)
Right, the NVCP perf setting you referenced directly affects GPU, not CPU clocks, so if you were experiencing that, the GPU must have been running at a very low clock to hold back average framerate like that.
I've only really seen anything like that in legacy 3D (think FFXIII-era) games that barely push modern GPUs.
Could be system and/or game-specific in your particular case, hard to say.
Crazyness wrote: ↑06 Sep 2022, 15:49
i cant remeber what i originaly had in mind with this post..sure i will remember and post again.
Sure, feel free to update this thread if/when you do remember.