Alternative to GSYNC-setting "VSync on + 142fps framelimit". What do you think?

What is the solution to this then if the framerate frequently dips? G-SYNC won't prevent the pre-render buffers from filling up right? Is this solved by setting maximum pre-rendered frames to 1?

 

Yeah, that's the reason why setting that to 1 helps prevent the issue.

In some cases, setting to 1 might result in a general performance hit. I haven't seen it myself, but other people have claimed some games get lower average FPS. Again, I can't confirm that myself.

 

borderless windowed , no v-sync , no screen tearing and no input lag ,
spared yourself 700 bucks for a stupid g-sync monitor with a chip that costs nvidia 50 bucks to make
my screen costs 130 bucks with 1 ms...
no ghosting , no screen tearing , no input delay...

never got why people pay 700-1000 bucks for a monitor while you can just put a game in bordered window and play without screen tearing.
got my projector via HDMI and dv-i for monitor , so i HAVE to play in bordered window on my monitor to play without screen tearing , dv-i always has screen tearing in fullscreen... projector cost me 50 bucks... watch movies and play games on 69'' screen , no screen tearing , not even with v-sync off , which is wierd i know , but projectors work differently i suppose... can play any game i want with this cheap thing ( even 4k dsr is no problem ) , never seen screen tearing

 

Troll attempt failed. Too obvious. You need more finesse.

 

I primarily test with CS:GO, so its setting of Max Prerendered Fames of 1 explains why there is negligible input lag even with Vsync

 

The input lag of CS:GO with vsync is not negligible. It's lower than other games, but far from negligible.

It would seem that you are just not sensitive to this. If I use vsync with CS:GO, the mouse feels like a boat at 60Hz.

 

Yes it's horrible at 60 Hz, but only barely detectible at 240 Hz. fps_max also behaves strangely. If you initially set it to 60, it lags. Then if you set it to 59 the lag is gone. But if you increase it again, you can go all the way up to 63 before it starts lagging again.

 

The 143/239 vs. 144/240 you are seeing on the readouts is a rounding error. 144Hz and 240Hz are actually 143.9/239.9Hz, give/take a few sub-decimal points.

There a couple of key differences between G-SYNC + V-SYNC and standalone V-SYNC when the maximum refresh rate is exceeded. Also, limiting the FPS 1 below the refresh rate is not enough to eliminate V-SYNC-level input latency with G-SYNC + V-SYNC.

I have to sit on my hands however until my complete "G-SYNC 101" article containing the answer to the above questions releases Monday on blurbusters.com.

Keep tuned.

 

I agree with mostly everything you say about this topic.

In my personal experience however, setting pre-rendered frames ('flip queue size' on my videocard) to 1 is not recommended on my (ancient) system (frame rate is capped slightly below refresh rate using RadeonPro, v-sync enabled).

The problem is that many games (I have more than 120 installed currently) start to stutter with flip queue size set to 1. In some cases it works fine and eliminates even more input lag, but in many cases the game just starts to exhibit small frame skips every now and then, sometimes even multiple in a row (the fps counter occasionally reflects this - a small, very quick drop in fps, and then instantly back up and stable again). The problem gets resolved completely by setting flip queue size to a value of 2, which is my safe minimum.

I'm not sure why this happens, but it definitely does, on two of my (very comparable) systems at least.

Specs are still actual (yeah, I should upgrade).

 

The majority of games I own are from 2007-2011 and run on DX9, some on OpenGL - which shows the same behaviour. The handful of games I own that run on DX10 or 11 don't seem to react to the 'flip queue size' setting, at least not that I can tell.

I'm playing on W7-64.

 

How about for those without any G-SYNC? Thanks.

 

What do you want to know?

 

Nothing else since I read what you told me, but since he's going to write the article so I suggest included for non G Sync too. Cheers.

 

In CS:GO at 60 Hz with G-SYNC + Vsync, setting fps_max to 59 does eliminate the horrendous lag at 60 fps. However limiting via RTSS only works at 58.

 

It features input latency tests that compare G-SYNC against V-SYNC OFF, double buffer V-SYNC, and Fast Sync across six refresh rates, including 240Hz, where I'll tell you now, almost anything is low latency if you can sustain that framerate.

The tests took weeks, so the article has been in the works for a while.

There's a difference between playing at 144Hz at 60 FPS, and playing at 60Hz at 60 FPS on a G-SYNC monitor; the first has (but for the increase in frametimes at that framerate) no additional latency, the second can have anywhere from 2-6 frames of additional delay.

 

I did not see this analysis here, but might help with regard to latency/input lag question.

Freesync vs Gsync - Delay analysis.


https://youtu.be/mVNRNOcLUuA

 

I am talking about forcing the monitor to 60 Hz.

At 60 Hz with G-SYNC + VSync, "fps_max 59" in CS:GO instantly removes the delay. It is somewhat inaccurate because if you start the game with "fps_max 60", it lags. If you then reduce it to 59 the lag is gone. If you then increase it back to 60, it does not lag. I must increase it all the way to 63 to make it lag again. However when reducing it, you must go down to 59 to make the lag stop again.

For an external frame limiter like RTSS, I need to set it to 58 to stop the lag.

 

CS:GO's frame limiter is not as accurate as RTSS. It seems to cap lower than what you set it to, and has unstable frame pacing. It is lag-free though, in contrast to RTSS.

 

If you're forcing the monitor to a max refresh rate of 60Hz via the Nvidia control panel before launching CS:GO (or using a launch command to control the refresh rate), then yes, you're effectively at native 60Hz, even on a 144Hz monitor. However, if you're max refresh rate is set to 144Hz, and your limiting the framerate to 60 FPS, the 144Hz scanout speed remains in full effect with G-SYNC, and no additional delay is introduced.

As RealNC stated, while latency-free, CS:GO's fps limiter is unstable, and can fluctuate/drift quite a lot, even when you set it to the wanted limit.

I won't post a direct link here, but my "G-SYNC 101" article is complete and now published on blurbusters.com. All the answer for this and more are there...

 

@jorimt I have read it. By the way, 2.1.3.7 is out.