GSync on or off (with Vsync disabled and high fps)

[Kaido]

I know, I know.. Another GSync settings thread

I've read the GSync 101 guide, and from it I understand that I should disable GSync to minimize input lag when playing with high (250+) fps and Vsync disabled. However, Battle(non)sense lists in one of his videos, that you should have GSync enabled.

So which one is it?

Is it monitor dependent? I'm talking about monitors with hardware GSync implementations here. I'm specifically interested in what is the case with Asus ROG PG278Q, since it's my current monitor.

I've come across claims of GSync mode adding ~2ms input lag in couple of different places. Could this actually be the case?

Some monitor review sites, TFT Central for example, test the input lag of the monitors with GSync enabled.

I hope someone can clear this out.

[jorimt]

I've read the GSync 101 guide, and from it I understand that I should disable GSync to minimize input lag when playing with high (250+) fps and Vsync disabled. However, Battle(non)sense lists in one of his videos, that you should have GSync enabled.

So which one is it?

Is it monitor dependent? I'm talking about monitors with hardware GSync implementations here.

I've come across claims of GSync mode adding ~2ms input lag in couple of different places. Could this actually be the case?

G-SYNC doesn't add input lag, G-SYNC prevents tearing, and a lack of tearing with G-SYNC + an FPS limit within the refresh rate, specifically, prevents sustained higher-than-refresh-rate frametime.

The more accurate way to express it, is that no sync can further reduce input lag.

As for why Battle(non)sense recommends G-SYNC over that scenario, it's all about context. The higher the max refresh rate, the faster the scanout cycle, and the faster the scanout cycle, the less difference there is between G-SYNC and no sync, thus, 500+ FPS no sync at 360Hz is much less beneficial from an input lag reduction standpoint than say 500+ FPS no sync at 60Hz.

It basically comes to a point of diminishing returns.

As for monitor-dependence, actual G-SYNC or max Hz operation will not change across monitors with equivalent specs, but individual monitor processing (and, to a lesser degree, GtG levels) can indirectly affect G-SYNC responsiveness (and all responsiveness) by simply lowering or raising the base minimum achievable input lag.

[Kaido]

Thank you for your reply!
Maybe I weren't specific enough or I just don't understand your answer thoroughly.

The scenario I am interested in is: playing with high stable fps, maximum monitor refresh rate, using in-game fps limiter, sometimes several thousand fps, VSync off.
So I'm not really using GSync, but I can still switch between the GSync-mode on and off for the monitor.

Did you ever test wheter simply enabling or disabling the GSync mode in the monitor makes any difference to input lag with high fps scenarios?

[jorimt]

The scenario I am interested in is: playing with high stable fps, using in-game fps limiter, sometimes several thousand fps, VSync off.
So I'm not really using GSync, but I can still switch between the GSync-mode on and off for the monitor.

The benefit of that depends, in part, on how high your max refresh rate is, and if your system is or isn't GPU-bound (GPU usage is maxed) at that framerate.

At 240Hz and higher max refresh rates with sustain framerates far above said refresh rate, your system, input devices, display processing, and the game engine become the bottleneck in minimum achievable input lag:

Did you ever test wheter simply enabling or disabling the GSync mode in the monitor makes any difference to input lag with high fps scenarios?

On a native G-SYNC monitor w/module, it shouldn't, and it only has an effect when the framerate is within the refresh rate and G-SYNC is engaged. E.g.:

G-SYNC on + V-SYNC off & G-SYNC off + V-SYNC off = standalone no sync with framerate sustained above the refresh rate.

[Kaido]

Thank you for the further clarification.

In that G-Sync vs. V-Sync OFF scenario, is G-Sync mode also disabled from NVidia control panel (nvcpl) when testing V-Sync OFF?

My subjective experiences in the couple of games I have 2000hrs+ on tell me that there might be something going on with the G-Sync mode. It just feels snappier when G-Sync is disabled from nvcpl, at least with my Asus pg278q at 144hz.

I was maybe hoping for some scientific results to clarify the situation since unfortunately I don't own the required equipment to test it myself.

By the way, for some reason my GPU runs considerably lower idle clocks on desktop when I enable G-Sync from nvcpl.
But that is probably just a bug. A bug which has been around for years tho..

[jorimt]

In that G-Sync vs. V-Sync OFF scenario, is G-Sync mode also disabled from NVidia control panel (nvcpl) when testing V-Sync OFF?

Yes.

My subjective experiences in the couple of games I have 2000hrs+ on tell me that there might be something going on with the G-Sync mode. It just feels snappier when G-Sync is disabled from nvcpl, at least with my Asus pg278q at 144hz.

I was maybe hoping for some scientific results to clarify the situation since unfortunately I don't own the required equipment to test it myself.

On a native G-SYNC monitor, no. In fact, if there were anything to this, it would be the opposite; monitor lag would possibly be lower with G-SYNC mode activated, as VRR requires a realtime scanout to function properly.

That, and on a native G-SYNC monitor, the signal runs through the module whether G-SYNC mode is enabled or disabled, as the module replaces the scaler on the monitor.

In other words, monitor processing latency shouldn't be affected on a native G-SYNC monitor whether you have the mode enabled/disabled.

By the way, for some reason my GPU runs considerably lower idle clocks on desktop when I enable G-Sync from nvcpl.
But that is probably just a bug. A bug which have been around for years tho..

Clock idling issues aren't exclusive to G-SYNC mode; there have indeed been desktop quirks involving GPU clocks and Nvidia drivers on and off for years now.

[Kaido]

On a native G-SYNC monitor, no. In fact, if there were anything to this, it would be the opposite; monitor lag would possibly be lower with G-SYNC mode activated, as VRR requires a realtime scanout to function properly.

That, and on a native G-SYNC monitor, the signal runs through the module whether G-SYNC mode is enabled or disabled, as the module replaces the scaler on the monitor.

In other words, monitor processing latency shouldn't be affected on a native G-SYNC monitor whether you have the mode enabled/disabled.

I see. That is interesting. It certainly shouldn't add any delay then, unless the tuned variable overdrive is so much more laid back in G-Sync mode that its causing some extra delay. But yea.. maybe not

I appreciate your time and effort in clearing this all out for me, thank you.

[slaver01]

In that G-Sync vs. V-Sync OFF scenario, is G-Sync mode also disabled from NVidia control panel (nvcpl) when testing V-Sync OFF?

Yes.

My subjective experiences in the couple of games I have 2000hrs+ on tell me that there might be something going on with the G-Sync mode. It just feels snappier when G-Sync is disabled from nvcpl, at least with my Asus pg278q at 144hz.

I was maybe hoping for some scientific results to clarify the situation since unfortunately I don't own the required equipment to test it myself.

On a native G-SYNC monitor, no. In fact, if there were anything to this, it would be the opposite; monitor lag would possibly be lower with G-SYNC mode activated, as VRR requires a realtime scanout to function properly.

That, and on a native G-SYNC monitor, the signal runs through the module whether G-SYNC mode is enabled or disabled, as the module replaces the scaler on the monitor.

In other words, monitor processing latency shouldn't be affected on a native G-SYNC monitor whether you have the mode enabled/disabled.

By the way, for some reason my GPU runs considerably lower idle clocks on desktop when I enable G-Sync from nvcpl.
But that is probably just a bug. A bug which have been around for years tho..

Clock idling issues aren't exclusive to G-SYNC mode; there have indeed been desktop quirks involving GPU clocks and Nvidia drivers on and off for years now.

Instead on a g-sync compatible monitor? (asus rog strix xg248q)

[jorimt]

I see. That is interesting. It certainly shouldn't add any delay then, unless the tuned variable overdrive is so much more laid back in G-Sync mode that its causing some extra delay. But yea.. maybe not

I appreciate your time and effort in clearing this all out for me, thank you.

You're welcome, and yes, GtG/overdrive won't increase/decrease input lag, as it occurs after the frame has been rendered and is being scanned into the display. Instead, it affects how blurry/clear pixel transitions are while frames are being scanned in.

In other words, on a monitor with slower GtG time, you're seeing the same information at the same time as a monitor with a faster GtG (assuming the Hz/max refresh rate are the same on both displays), it will just smear/ghost more.

Instead on a g-sync compatible monitor? (asus rog strix xg248q)

I've only tested on native G-SYNC monitors, so I can only speak to those. G-SYNC Compatible monitors don't have a module, so they are beholden to the driver only.

It's possible some G-SYNC Compatible monitors may have lower monitor processing latency with G-SYNC mode enabled. It could vary depending on how the processing is handled on the monitor.

Also, somewhat unrelated, but I know that fixed overdrive presets sometimes behave differently on G-SYNC Compatible monitors depending on if G-SYNC mode is on or off.