i have a pc with a 9700x and a 5070
with comp settings my game can get well over 400 frames in game
my monitor is only 180hz though
has anyone tested the latency with gsync and a 175 fps cap compared to nvidia fast sync running at 300+ fps using LDAT?
has anyone actually tried fast sync with LDAT in recent years?
Personally I use Gsync + RTSS cap with reflex limiter @171fps (I have a 180hz monitor) + no reflex command
Also no vsync and no low latency mode in nvcp
Fast Sync is triple buffering and incurs a higher latency penalty than G-Sync + V-Sync + framerate cap (Reflex / framerate limiter) and, on top of that, has uneven frame delivery due to the refresh rate not being synchronized to the framerate.
There’s no point in Fast Sync when you have G-Sync unless G-Sync itself causes you other issues (such as OLED flicker).
fast sync is different than triple buffering if im not mistaken but ill have to look that up
Fast Sync is true triple buffering, the sort that allows the framerate to exceed the refresh rate by displaying the latest frame when a refresh starts and dropping older ones (scanout, last rendered, rendering are the 3 buffers).
The triple buffering we used to see in games uses a queue which adds 1 frame of input latency.
Never use gsync in cs, unless you’re a casual player
Either way you’re still going to suck
Don’t use any of that crap. Just cap frames at a high number if you want to and roll from there.
Fast Sync is triple buffering
that's bullshit
there's no 3rd buffer, just two "uroborus" circular buffers and a fancy "gear switch" = flip logic to allow presenting frames from either of them at any given time and then designating = tagging = naming it as "last rendered buffer" so the next one goes in the other one and so on, no such thing as 1-frame latency penalty - it's quite close to vsync off
g-sync is also double-buffered so it has similar latency (both are in the .5ms range)
g-sync is just smoother (vsync fast suffers from some stutters in wide panning if the fps is not drastically over refresh, like 4x-5x)
however at low refresh rates g-sync sucks ballz for any kind of competitive shooter since the input lag when capped at let's say 144-180 fps is much higher than if running at full potential of let's say 600+ fps
the ideal is to have neither i.e. a very high refresh rate monitor 480hz+ and then up res & video settings / cap with fps_max to never even reach it making screen tearing a non-issue
This is the best for smoothness but it incurs a latency penalty
>that's bullshit
there's no 3rd buffer, just two "uroborus" circular buffers and a fancy "gear switch" = flip logic to allow presenting frames from either of them at any given time and then designating = tagging = naming it as "last rendered buffer" so the next one goes in the other one and so on
From Nvidia's own explanation of what Fast Sync is:
3 buffers:
1) Front buffer for scan-out
2) Back buffer for rendering
3) Last rendered buffer to store the last frame rendered
Without that 3rd buffer, you fall back to regular V-Sync behavior.
Last time I checked, 3 is "triple" in the English language and so this is "triple buffering". So, before you rudely call bullshit on an Internet stranger's response, make sure you at least have the right information.
>no such thing as 1-frame latency penalty - it's quite close to vsync off
You're responding to a claim I didn't make. I was referring to triple buffering as implemented in games - a frame queue - not to Fast Sync.
>however at low refresh rates g-sync sucks ballz for any kind of competitive shooter since the input lag when capped at let's say 144-180 fps is much higher than if running at full potential of let's say 600+ fps
This has been demonstrated to be false several times, starting with Blur Buster's findings almost a decade ago. G-Sync at even lower refresh rates like 144Hz adds \~1ms input latency over the best case input latency with everything off (V-Sync, G-Sync, high framerate limit), while giving you complete frames instead of torn frames (which makes the benefit of torn frames that arrive a tiny bit sooner due to tearing questionable). Without a framerate limit, the GPU is pegged at \~100% usage and this causes additional input latency \again\, completely negating the \~1ms penalty incurred by G-Sync + V-Sync + framerate cap below refresh rate and adding on top of that even more.
At higher refresh rates the comparison becomes completely senseless and you should always, always use G-Sync + V-Sync + Reflex. I don't care what CS2 pros who run their monitors at 1280x960 stretched with 400-800 DPI on a modern gaming mouse say, this is just what they're used to from playing on potato PCs when they got started, there's nothing scientific about it.
Where did you get that information from?
Check my reply to another comment that replied to this comment below.