FreeSync vs. G-Sync Compatible: Unexpected Input Lag Results

Megalith

Megalith

24-bit/48kHz
Staff member
Joined
Aug 20, 2006
Messages
13,000
YouTuber Battle(non)sense decided to test whether there were any differences between running an ASUS VG258Q monitor in FreeSync mode vs. G-Sync mode on NVIDIA and AMD cards. They appear to perform quite comparably, although certain combinations (e.g., AMD GPU + FreeSync) appear to offer lower delay. One interesting find was input lag of 10ms or more with NVIDIA when adaptive sync was turned off entirely.

You can now use an adaptive-sync aka. FreeSync monitor on an Nvidia graphics card without any workarounds or hacks as the Nvidia driver now comes with native adaptive-sync support. So I set out to test if there is any difference between running a monitor in FreeSync, or G-Sync Compatible mode. What did look like a fast and straight forward test turned out to be quite a challenge which lead to some interesting results and raised more questions about adaptive sync monitors on Nvidia graphics cards.
 
Interesting results. I knew vsync caused input lag, but had no idea limiting frame rates did. Also quite bizarre that enabling adaptive sync on an Nvidia card and freesync monitor reduced input lag.
 
Brian_B said:
Is there a TL;DR for those of us that can't (or prefer not to) watch the video?
Click to expand...

He used CS: GO at 1080p set to the lowest setting for his tests

RX 580:
No Freesync and no Vsync- Game running at 300 fps at lowest settings Average delay 16.3ms
Vsync on- Game running at 144 fps Average delay 42.73ms (about 6 frames at 144hz)
In game frame limiter set to 138 fps Average delay 19.8ms
Using AMD's Frame-Rate Target Control feature in AMD drivers at 138 fps Average delay 42.47ms (yikes)
Rivatuner fps limiter set to 138fps Average Delay 25.2ms
Freesync on no vsync in-game fps cap of 138 Average Delay 20.73ms (not a big change compared to using the in-game limiter by itself)
Freesync with FRTC at 138 Average Delay 41.93ms
Freesync with Rivatuner limit set to 138 Average Delay 24.47ms
Freesync + Vsync + In-game FPS limit set at 138 Average Delay 21.27ms
Freesync + Vsync with no cap Average Delay 43.67

GTX 1060
No Gsync no Vsync and no in-game cap Average Delay 25.33ms (He tried different cards, different systems, brand new Win10 installs, disabling the Freesync setting in the monitor itself, results stayed the same)
Vsync on no in-game cap Average Delay 58.47ms
No Vsync in-game fps capped at 138 Average Delay 26.33ms
No Vysnc Rivatuner limited 138 Average Delay 30.67ms
Gsync On in-game limit of 138 Average Delay 18.87ms
Gsync On Rivatuner limit of 138 Average Delay 25.2
Gsync and Vsync on with an in-game limit of 138 Average Delay 20.6ms
Gsync and Vsync on with no fps cap Average Delay 49.47ms

Random notes:
With this monitor sometimes enabling the Adapative Sync setting in the monitor would cause half the screen to go black. He reached out to Nvidia and they were able to reproduce the issue in-house and are working on a fix. Turning the monitor off and on again would allow the feature to work.
He told Nvidia about the delay disparity this monitor shows between AMD and Nvidia cards, but has not yet heard back from them about it.
He plans to get a different monitor and see if the disparity exists across the board or if it is specific to this one monitor.
 
JosiahBradley said:
TL:DR AMD cards have less input lag with adaptive sync off. G-sync and freesync about equal and lower for input lag. Make sure to keep your frame rate below the refresh rate if you plan on turning on vsync.
Click to expand...

How much below? Is there a magic formula for this?
 
GSDragoon said:
How much below? Is there a magic formula for this?
Click to expand...

Around 4-5% below I'd say. So the 138 the guy from the video used is a good number for 144hz monitors. For 120hz maybe somewhere in the 115-ish range. At 60hz probably no lower than 55.
 
Derangel said:
He used CS: GO at 1080p set to the lowest setting for his tests

RX 580:
No Freesync and no Vsync- Game running at 300 fps at lowest settings Average delay 16.3ms
Vsync on- Game running at 144 fps Average delay 42.73ms (about 6 frames at 144hz)
In game frame limiter set to 138 fps Average delay 19.8ms
Using AMD's Frame-Rate Target Control feature in AMD drivers at 138 fps Average delay 42.47ms (yikes)
Rivatuner fps limiter set to 138fps Average Delay 25.2ms
Freesync on no vsync in-game fps cap of 138 Average Delay 20.73ms (not a big change compared to using the in-game limiter by itself)
Freesync with FRTC at 138 Average Delay 41.93ms
Freesync with Rivatuner limit set to 138 Average Delay 24.47ms
Freesync + Vsync + In-game FPS limit set at 138 Average Delay 21.27ms
Freesync + Vsync with no cap Average Delay 43.67

GTX 1060
No Gsync no Vsync and no in-game cap Average Delay 25.33ms (He tried different cards, different systems, brand new Win10 installs, disabling the Freesync setting in the monitor itself, results stayed the same)
Vsync on no in-game cap Average Delay 58.47ms
No Vsync in-game fps capped at 138 Average Delay 26.33ms
No Vysnc Rivatuner limited 138 Average Delay 30.67ms
Gsync On in-game limit of 138 Average Delay 18.87ms
Gsync On Rivatuner limit of 138 Average Delay 25.2
Gsync and Vsync on with an in-game limit of 138 Average Delay 20.6ms
Gsync and Vsync on with no fps cap Average Delay 49.47ms

Random notes:
With this monitor sometimes enabling the Adapative Sync setting in the monitor would cause half the screen to go black. He reached out to Nvidia and they were able to reproduce the issue in-house and are working on a fix. Turning the monitor off and on again would allow the feature to work.
He told Nvidia about the delay disparity this monitor shows between AMD and Nvidia cards, but has not yet heard back from them about it.
He plans to get a different monitor and see if the disparity exists across the board or if it is specific to this one monitor.
Click to expand...
Did he report both issues separatly, or together?
 
Wow, I never spend over 5 minutes on a single article, so video "articles" are definitely not with me. I clicked near the end just out of curiosity, and I hear him say "if you have an Nvidia card and want an adaptive/freesync monitor, I suggest you buy a G-sync compatible one, because while non-compatible might work, but as we saw even G-sync compatible ones can still have issues" ??? WTF logic is that? How do you get that conclusion (from a single data point no less!)?
 
ecuador said:
Wow, I never spend over 5 minutes on a single article, so video "articles" are definitely not with me. I clicked near the end just out of curiosity, and I hear him say "if you have an Nvidia card and want an adaptive/freesync monitor, I suggest you buy a G-sync compatible one, because while non-compatible might work, but as we saw even G-sync compatible ones can still have issues" ??? WTF logic is that? How do you get that conclusion (from a single data point no less!)?
Click to expand...

You'd probably understand how he got to that point if you didn't just skip to the end of his 17 minute video.
 
ecuador said:
Wow, I never spend over 5 minutes on a single article, so video "articles" are definitely not with me. I clicked near the end just out of curiosity, and I hear him say "if you have an Nvidia card and want an adaptive/freesync monitor, I suggest you buy a G-sync compatible one, because while non-compatible might work, but as we saw even G-sync compatible ones can still have issues" ??? WTF logic is that? How do you get that conclusion (from a single data point no less!)?
Click to expand...
Note that "g-sync compatible" includes (or rather, is limited to) freesync displays that nvidia certified as compatible with g-sync.
 
Derangel said:
You'd probably understand how he got to that point if you didn't just skip to the end of his 17 minute video.
Click to expand...

The logic is "data point from group A which nVidia guarantees has issues, so get that, and avoid group B which we did not test at all". I certainly would not waste 17 minutes to see how they make that logic leap.
 
ecuador said:
The logic is "data point from group A which nVidia guarantees has issues, so get that, and avoid group B which we did not test at all". I certainly would not waste 17 minutes to see how they make that logic leap.
Click to expand...

A monitor certified by Nvidia to be compatible has serious issues. I don't entirely agree with him, but if a certified monitor has a major issue that Nvidia was even able to reproduce in-house in means it isn't restricted just to his monitor. So warning people that monitors could have issues is not unfounded. Based on the beginning of his video I feel like his audience might not be as fully informed on this stuff as people from a tech enthusiast site.
 
IdiotInCharge said:
Means 'least amount of issues'. If Nvidia held Freesync monitors up to the G-Sync standard, none would pass.
Click to expand...

I'd imagine "half the monitor goes black when enabling adaptive synch" should also be something that disqualifies it. If that's something that wasn't caught in testing its a sign that other certified monitors could have issues like that.
 
Derangel said:
I'd imagine "half the monitor goes black when enabling adaptive synch" should also be something that disqualifies it. If that's something that wasn't caught in testing its a sign that other certified monitors could have issues like that.
Click to expand...

Or sample variation. Could be a corner case with the OP's setup. Who knows with the broad spectrum of Freesync monitors out there.
 
IdiotInCharge said:
Or sample variation. Could be a corner case with the OP's setup. Who knows with the broad spectrum of Freesync monitors out there.
Click to expand...

Nvidia reproduced the error in-house. So its likely something with that monitor on Nvidia cards or a driver bug.
 
So... Nvidia says freesync is shit becuase nvidia+ freesync is shit.. is this what we learned here?
 
Uvaman2 said:
So... Nvidia says freesync is shit becuase nvidia+ freesync is shit.. is this what we learned here?
Click to expand...

Nope. Nvidia + Async/Freesync has less (marginally) input lag than AMD + Async/freesync. The issue with this monitor (and it may or may not be limited to this monitor) is that the lag is much higher with async turned off on an Nvidia card vs an AMD one. Plus the black screen issue that also may or may not be something specific to this monitor with an Nvidia card and something Nvidia is working on a fix for.
 
PhaseNoise said:
My god I hate the new world order of article-in-a-video.

Charts good. Being able to link to a page, good. Table of contents so you can go where you want, good.
Click to expand...

Same, an article i can read in 5 minutes vs a long video, its just laziness at the end of the day on there part, they could just put the script up as an article, they have already written that for the video..
 
Dodge245 said:
Same, an article i can read in 5 minutes vs a long video, its just laziness at the end of the day on there part, they could just put the script up as an article, they have already written that for the video..
Click to expand...

So likely spending several hours making a video is more lazy than turning a script into a written article? Your logic is quite lacking here.
 
Derangel said:
So likely spending several hours making a video is more lazy than turning a script into a written article? Your logic is quite lacking here.
Click to expand...

i think you misunderstood what i was trying to say.. its lazy of them to not just post the script used in the video online so i can choose whether to watch or read. The script doesn't need to be a full blown article, just post the script, that would be fine with me.
 
Derangel said:
A monitor certified by Nvidia to be compatible has serious issues. I don't entirely agree with him, but if a certified monitor has a major issue that Nvidia was even able to reproduce in-house in means it isn't restricted just to his monitor. So warning people that monitors could have issues is not unfounded. Based on the beginning of his video I feel like his audience might not be as fully informed on this stuff as people from a tech enthusiast site.
Click to expand...
But he doesn't make the point of, hey hold off your buying this monitor or even a gsync monitor until we investigate these issues, or just a caveat emptor in general for people buying one. Instead he says do buy one of these monitors, but don't buy the other ones we did not test at all because they could be worse... If you don't see the logic leap there...
 
Can a video lover create a cliff note narrative webpage with references and graphs please? Thanks.
 
Derangel said:
He used CS: GO at 1080p set to the lowest setting for his tests

RX 580:
No Freesync and no Vsync- Game running at 300 fps at lowest settings Average delay 16.3ms
Vsync on- Game running at 144 fps Average delay 42.73ms (about 6 frames at 144hz)
In game frame limiter set to 138 fps Average delay 19.8ms
Using AMD's Frame-Rate Target Control feature in AMD drivers at 138 fps Average delay 42.47ms (yikes)
Rivatuner fps limiter set to 138fps Average Delay 25.2ms
Freesync on no vsync in-game fps cap of 138 Average Delay 20.73ms (not a big change compared to using the in-game limiter by itself)
Freesync with FRTC at 138 Average Delay 41.93ms
Freesync with Rivatuner limit set to 138 Average Delay 24.47ms
Freesync + Vsync + In-game FPS limit set at 138 Average Delay 21.27ms
Freesync + Vsync with no cap Average Delay 43.67

GTX 1060
No Gsync no Vsync and no in-game cap Average Delay 25.33ms (He tried different cards, different systems, brand new Win10 installs, disabling the Freesync setting in the monitor itself, results stayed the same)
Vsync on no in-game cap Average Delay 58.47ms
No Vsync in-game fps capped at 138 Average Delay 26.33ms
No Vysnc Rivatuner limited 138 Average Delay 30.67ms
Gsync On in-game limit of 138 Average Delay 18.87ms
Gsync On Rivatuner limit of 138 Average Delay 25.2
Gsync and Vsync on with an in-game limit of 138 Average Delay 20.6ms
Gsync and Vsync on with no fps cap Average Delay 49.47ms

Random notes:
With this monitor sometimes enabling the Adapative Sync setting in the monitor would cause half the screen to go black. He reached out to Nvidia and they were able to reproduce the issue in-house and are working on a fix. Turning the monitor off and on again would allow the feature to work.
He told Nvidia about the delay disparity this monitor shows between AMD and Nvidia cards, but has not yet heard back from them about it.
He plans to get a different monitor and see if the disparity exists across the board or if it is specific to this one monitor.
Click to expand...
Wow, that saved me 16 minutes thank you.. I think the other issue is people don't want to host their own reviews.
 
Mega6 said:
Can a video lover create a cliff note narrative webpage with references and graphs please? Thanks.
Click to expand...
Read derangels post ya lazy fuck :p. No one is going to cap graphs for you though lol. I can see there is a market for this as the shitty video rambling reviews continue to increase.
 
  • Like
Reactions: Mega6
like this
N4CR said:
Read derangels post ya lazy fuck :p. No one is going to cap graphs for you though lol. I can see there is a market for this as the shitty video rambling reviews continue to increase.
Click to expand...
No, I am a lazy fake and u can't make me. Excuse me while my AAD kicks in, sorry didn't even finish reading your posat either.

Should have been more diligent and put "+1 or x2".

Sorry
 
  • Like
Reactions: N4CR
like this
PhaseNoise said:
My god I hate the new world order of article-in-a-video.

Charts good. Being able to link to a page, good. Table of contents so you can go where you want, good.
Click to expand...

But you know who doesn’t hate this new world order? Advertisers!
 
N4CR said:
Read derangels post ya lazy fuck :p. No one is going to cap graphs for you though lol. I can see there is a market for this as the shitty video rambling reviews continue to increase.
Click to expand...

Q0SEx8F.png
 
did a quick video. freesync monitor on gtx1070


not very scientific but gsync looks to be working on freesync display
 
You must log in or register to reply here.
Back
Top