[Thread Superseded] G-Sync 101 w/Chart (WIP)

Talk about NVIDIA G-SYNC, a variable refresh rate (VRR) technology. G-SYNC eliminates stutters, tearing, and reduces input lag. List of G-SYNC Monitors.
Locked
User avatar
jorimt
Posts: 2481
Joined: 04 Nov 2016, 10:44
Location: USA

Re: G-Sync 101 w/Chart (WIP)

Post by jorimt » 16 Mar 2017, 19:14

akirru wrote:I would like to start by saying thank-you for all of your hard work jorimt :D

I just had a quick question to ask you. Does it matter at all if you change the pre-rendered frames whilst gsync is activated? I'm quite sensitive to mouse lag. And I usually set it to pre-rendered frame "1" ... I just wanted to find out if its recommended to do this whilst using G-sync.

Thanks
Andrew
I'm glad you found my OP useful (more to come).

I actually addressed that question a few pages ago:
http://forums.blurbusters.com/viewtopic ... =80#p24438
(jorimt: /jor-uhm-tee/)
Author: Blur Busters "G-SYNC 101" Series

Displays: ASUS PG27AQN, LG 48CX VR: Beyond, Quest 3, Reverb G2, Index OS: Windows 11 Pro Case: Fractal Design Torrent PSU: Seasonic PRIME TX-1000 MB: ASUS Z790 Hero CPU: Intel i9-13900k w/Noctua NH-U12A GPU: GIGABYTE RTX 4090 GAMING OC RAM: 32GB G.SKILL Trident Z5 DDR5 6400MHz CL32 SSDs: 2TB WD_BLACK SN850 (OS), 4TB WD_BLACK SN850X (Games) Keyboards: Wooting 60HE, Logitech G915 TKL Mice: Razer Viper Mini SE, Razer Viper 8kHz Sound: Creative Sound Blaster Katana V2 (speakers/amp/DAC), AFUL Performer 8 (IEMs)

MT_
Posts: 113
Joined: 17 Jan 2017, 15:39

Re: G-Sync 101 w/Chart (WIP)

Post by MT_ » 18 Mar 2017, 07:22

Perhaps useful to some, my findings anyway, this could deviate from game to game.

HPET here keeps my internal game frame limiter more stable, versus pure TSC. So less fluctuations and more stable frametimes.
Assuming you do not want to use V-sync, otherwise you get frametime stabilization anyway. But even with V-sync on, it could mean hitting the V-sync barrier or not.

This could be handy in case you want to maximize your FPS and keep G-sync bottom tearing mostly stable.

HPET vs TSC also affect performance as my tests shows:

_____________________________

HPET Off, UsePlatformClock False
299 fps

HPET On, UsePlatformClock False
299 fps

HPET On, UsePlatformClock True
275 fps

HPET Off, UsePlatformClock True
275 fps

This is on maximized CPU load and minimal GPU load.
It seems as only when you force HPET through Windows as well, FPS drops start to happen.

___________________________

Perhaps a bit irrelevant info but oh well. :P

Also tweaking Windows timer resolution from 0.1000ms to 0.500ms can drift in-game FPS cap up or downwards depending on which hardware timer used, so for some additional finetuning perhaps.

Cheers!
LTSC 21H2 Post-install Script
https://github.com/Marctraider/LiveScript-LTSC-21H2

System: MSI Z390 MEG Ace - 2080 Super (300W mod) - 9900K 5GHz Fixed Core (De-lid) - 32GB DDR3-3733-CL18 - Xonar Essence STX II

akirru
Posts: 7
Joined: 15 Mar 2017, 08:23

Re: G-Sync 101 w/Chart (WIP)

Post by akirru » 20 Mar 2017, 07:21

jorimt wrote:
akirru wrote:I would like to start by saying thank-you for all of your hard work jorimt :D

I just had a quick question to ask you. Does it matter at all if you change the pre-rendered frames whilst gsync is activated? I'm quite sensitive to mouse lag. And I usually set it to pre-rendered frame "1" ... I just wanted to find out if its recommended to do this whilst using G-sync.

Thanks
Andrew
I'm glad you found my OP useful (more to come).

I actually addressed that question a few pages ago:
http://forums.blurbusters.com/viewtopic ... =80#p24438
Thanks for getting back to me jorimt :)

Interesting and as you said I think it is an inconsistent variable. I get worse performance in some games and still no noticeable change in input lag. I certainly think capping the framerate below the threshold of G-sync might negate any real input lag anyway. I have it capped at "141" on Rivatuner and most games feel butter smooth. I mean I still get increased lag when I have a massive drop in fps. But that isn't the fault of G-sync.

EDIT: Something I have noticed but not really thoroughly looked into. Is the effect your mouse can have on input lag and Gsync itself. For example I have a g400s which stutters a lot. And a rat5 (pretty bad for competitive gaming as it has a laser) which is really smooth. But overall the g400s has less input lag.

capcomasd
Posts: 2
Joined: 20 Mar 2017, 08:34

Re: G-Sync 101 w/Chart (WIP)

Post by capcomasd » 20 Mar 2017, 08:58

Excelent post jorimt, you've answered many doubts to many gamers like me. However, after reading all this post I still have a question: Having G-SYNC Activated + V-SYNC ON in NVIDIA CONTROL PANEL, I should turn ON or OFF V SYNC IN-GAME? Does it matter for example if I have V SYNC ON in NVIDIA CONTROL PANEL and OFF IN-GAME?

By the way I play mostly Battlefield 1 in a Dell S2716DG G-SYNC monitor and GTX 1080.

User avatar
RealNC
Site Admin
Posts: 3737
Joined: 24 Dec 2013, 18:32
Contact:

Re: G-Sync 101 w/Chart (WIP)

Post by RealNC » 20 Mar 2017, 13:33

capcomasd wrote:Having G-SYNC Activated + V-SYNC ON in NVIDIA CONTROL PANEL, I should turn ON or OFF V SYNC IN-GAME?
In-game vsync must be OFF. It can interfere with g-sync otherwise (like the game doing triple buffering, or using timing that assumes the rendering is synced to a fixed refresh, or doing an automatic 60FPS cap, or whatever.)

So, always OFF.

In games that don't have an in-game vsync setting you usually need to tweak some ini file.
SteamGitHubStack Overflow
The views and opinions expressed in my posts are my own and do not necessarily reflect the official policy or position of Blur Busters.

Bouttime
Posts: 11
Joined: 15 Jan 2017, 11:54

Re: G-Sync 101 w/Chart (WIP)

Post by Bouttime » 20 Mar 2017, 15:03

RealNC wrote:
capcomasd wrote:Having G-SYNC Activated + V-SYNC ON in NVIDIA CONTROL PANEL, I should turn ON or OFF V SYNC IN-GAME?
In-game vsync must be OFF. It can interfere with g-sync otherwise (like the game doing triple buffering, or using timing that assumes the rendering is synced to a fixed refresh, or doing an automatic 60FPS cap, or whatever.)

So, always OFF.

In games that don't have an in-game vsync setting you usually need to tweak some ini file.
Don't want to confuse things but........I thought the message from the OP was in NCP have G-SYNC on and V-SYNC 'use 3D application setting' and in game have V-SYNC On.

Or.....

In NCP have G-SYNC On and V-SYNC On and in game have V-SYNC Off.

Sent from my Nexus 6 using Tapatalk

User avatar
jorimt
Posts: 2481
Joined: 04 Nov 2016, 10:44
Location: USA

Re: G-Sync 101 w/Chart (WIP)

Post by jorimt » 20 Mar 2017, 15:22

akirru wrote: Thanks for getting back to me jorimt :)

Interesting and as you said I think it is an inconsistent variable. I get worse performance in some games and still no noticeable change in input lag. I certainly think capping the framerate below the threshold of G-sync might negate any real input lag anyway. I have it capped at "141" on Rivatuner and most games feel butter smooth. I mean I still get increased lag when I have a massive drop in fps. But that isn't the fault of G-sync.

EDIT: Something I have noticed but not really thoroughly looked into. Is the effect your mouse can have on input lag and Gsync itself. For example I have a g400s which stutters a lot. And a rat5 (pretty bad for competitive gaming as it has a laser) which is really smooth. But overall the g400s has less input lag.
Yes, as my input latency tests show, a 2-3 frame cap below your max refresh rate is enough to prevent additonal input latency with G-Sync + v-sync on. As stated in an edit on my OP, however, RTSS likely adds 1 frame of input latency over in-game caps, which add none. So in-game framerate limiters are still preferred.

I will be retesting for input latency across a wide range of scenarios in the coming weeks, and with improved testing equipment, so stay tuned.

Regarding frametime spikes (aka "massive fps drops"), with standalone double-buffer v-sync (G-Sync disabled), there will be long halts during frametime spikes, and with G-Sync + v-sync on, you'll get shorter halts, but they will still be there, as the frametime compensation mechanism suspends the affect frame just long enough to prevent tearing. With G-Sync + v-sync off/standlone v-sync off, you're going to see a complete middle tearline, but in most instance, with less/shorter halts.

As for the difference in input latency between your two mice, it's likely the polling rate. For high refresh rates and G-Sync, the higher the polling rate, the better. If either of your mice has a "1000 Hz" polling rate option, you'll need to enable that for the lowest input latency and optimal response.
capcomasd wrote:Excelent post jorimt, you've answered many doubts to many gamers like me. However, after reading all this post I still have a question: Having G-SYNC Activated + V-SYNC ON in NVIDIA CONTROL PANEL, I should turn ON or OFF V SYNC IN-GAME? Does it matter for example if I have V SYNC ON in NVIDIA CONTROL PANEL and OFF IN-GAME?

By the way I play mostly Battlefield 1 in a Dell S2716DG G-SYNC monitor and GTX 1080.
As RealNC stated, Nvidia Control Panel v-sync is the safest bet.

However, as I noted in my OP, there are (extremely) rare occasions where only the in-game v-sync setting will work for a given game, in which case you should enable it in-game instead.

An easy way to tell if the Nvidia Control Panel v-sync setting isn't working, is if you see tearing at any point with G-Sync enabled.

But yes, 99% of the time for G-Sync: v-sync on in Nvidia Control Panel, v-sync off in-game.
(jorimt: /jor-uhm-tee/)
Author: Blur Busters "G-SYNC 101" Series

Displays: ASUS PG27AQN, LG 48CX VR: Beyond, Quest 3, Reverb G2, Index OS: Windows 11 Pro Case: Fractal Design Torrent PSU: Seasonic PRIME TX-1000 MB: ASUS Z790 Hero CPU: Intel i9-13900k w/Noctua NH-U12A GPU: GIGABYTE RTX 4090 GAMING OC RAM: 32GB G.SKILL Trident Z5 DDR5 6400MHz CL32 SSDs: 2TB WD_BLACK SN850 (OS), 4TB WD_BLACK SN850X (Games) Keyboards: Wooting 60HE, Logitech G915 TKL Mice: Razer Viper Mini SE, Razer Viper 8kHz Sound: Creative Sound Blaster Katana V2 (speakers/amp/DAC), AFUL Performer 8 (IEMs)

capcomasd
Posts: 2
Joined: 20 Mar 2017, 08:34

Re: G-Sync 101 w/Chart (WIP)

Post by capcomasd » 20 Mar 2017, 15:30

jorimt wrote:
akirru wrote: Thanks for getting back to me jorimt :)

Interesting and as you said I think it is an inconsistent variable. I get worse performance in some games and still no noticeable change in input lag. I certainly think capping the framerate below the threshold of G-sync might negate any real input lag anyway. I have it capped at "141" on Rivatuner and most games feel butter smooth. I mean I still get increased lag when I have a massive drop in fps. But that isn't the fault of G-sync.

EDIT: Something I have noticed but not really thoroughly looked into. Is the effect your mouse can have on input lag and Gsync itself. For example I have a g400s which stutters a lot. And a rat5 (pretty bad for competitive gaming as it has a laser) which is really smooth. But overall the g400s has less input lag.
Yes, as my input latency tests show, a 2-3 frame cap below your max refresh rate is enough to prevent additonal input latency with G-Sync + v-sync on. As stated in an edit on my OP, however, RTSS likely adds 1 frame of input latency over in-game caps, which add none. So in-game framerate limiters are still preferred.

I will be retesting for input latency across a wide range of scenarios in the coming weeks, and with improved testing equipment, so stay tuned.

Regarding frametime spikes (aka "massive fps drops"), with standalone double-buffer v-sync (G-Sync disabled), there will be long halts during frametime spikes, and with G-Sync + v-sync on, you'll get shorter halts, but they will still be there, as the frametime compensation mechanism suspends the affect frame just long enough to prevent tearing. With G-Sync + v-sync off/standlone v-sync off, you're going to see a complete middle tearline, but in most instance, with less/shorter halts.

As for the difference in input latency between your two mice, it's likely the polling rate. For high refresh rates and G-Sync, the higher the polling rate, the better. If either of your mice has a "1000 Hz" polling rate option, you'll need to enable that for the lowest input latency and optimal response.
capcomasd wrote:Excelent post jorimt, you've answered many doubts to many gamers like me. However, after reading all this post I still have a question: Having G-SYNC Activated + V-SYNC ON in NVIDIA CONTROL PANEL, I should turn ON or OFF V SYNC IN-GAME? Does it matter for example if I have V SYNC ON in NVIDIA CONTROL PANEL and OFF IN-GAME?

By the way I play mostly Battlefield 1 in a Dell S2716DG G-SYNC monitor and GTX 1080.
As RealNC stated, Nvidia Control Panel v-sync is the safest bet.

However, as I noted in my OP, there are (extremely) rare occasions where only the in-game v-sync setting will work for a given game, in which case you should enable it in-game instead.

An easy way to tell if the Nvidia Control Panel v-sync setting isn't working, is if you see tearing at any point with G-Sync enabled.

But yes, 99% of the time for G-Sync: v-sync on in Nvidia Control Panel, v-sync off in-game.
Clear enough, thanks for the fast reply guys!

User avatar
RealNC
Site Admin
Posts: 3737
Joined: 24 Dec 2013, 18:32
Contact:

Re: G-Sync 101 w/Chart (WIP)

Post by RealNC » 22 Mar 2017, 03:55

I just tried the "max pre-rendered frames" setting with G-Sync in Witcher 3. I've set it to 8 in Inspector. Two settings need to be changed, because Inspector does not sync them. In the "Sync and Refresh" section "Maximum pre-rendered frames" must be set to 8, and in the "Other" section (scroll down to the bottom) "Maximum frames allowed" must be "0x00000008".

It makes a MASSIVE difference when no frame capping is done. Input lag is so big I can count tenths of seconds visually. It feels like it's somewhere around 200ms. It's borderline unplayable.

However, when capping the frame rate the input lag disappears. This is true for both the in-game capper as well as RTSS. However, when the game doesn't hit the cap (for example when capping to 80FPS but the game only puts out 70FPS), the huge input lag is back.

You can use custom frame rates to cap at with the in-game limiter by editing the "user.settings" file (it's in the "Documents\The Witcher 3" folder. Example:

LimitFPS=80

So at least for Witcher 3, it makes a TON of difference when either not capping the frame rate, or when the game's frame rate falls below the value you capped it at. Which for g-sync means setting pre-rendered frames to 1 is still best for input lag, since the whole point of g-sync is to allow the game to fall below the cap without causing stutters.

There was no performance or "smoothness" difference between a max frames setting of 1 and 8. Performance was exactly the same and no "hiccups" of any kind were there with a setting of 1.

As a side-note, it feels like RTSS adds a bit more input lag compared to the in-game limiter. I have no equipment to test this though. My way of testing was to disable the "hardware cursor" in the game's video settings and compare the in-game mouse cursor movement (not the camera pan, I mean the actual mouse cursor in the game's ESC menu.) With RTSS it does feel a bit "floaty", while with the in-game limiter it doesn't.
SteamGitHubStack Overflow
The views and opinions expressed in my posts are my own and do not necessarily reflect the official policy or position of Blur Busters.

MaxiJazz
Posts: 16
Joined: 14 Mar 2017, 05:13

Re: G-Sync 101 w/Chart (WIP)

Post by MaxiJazz » 22 Mar 2017, 07:44

"Maximum frames allowed" working only in OpenGL apps (c) Guru3d

My problem is DIABLO III - This is f*ing mess... G-sync, V-sync, Frame cap... This game is don't give a sh*t to this. It's stutter in town even with SOLID caps like 60\120\144 fps. (Game on SSD, but tried WD Black HDD and my USB 3.0 external HDD)

Locked