NVIDIA G-Sync Pulsar monitor - Asus ROG Strix XG27AQNGV

[jorimt]

Low-Latency Mode='On' (older Max-Prerendered Frames=1) is functionally identical to Nv Reflex, but it isn't aware of the internal game pipeline.

Reflex, unlike LLM "On," uses a dynamic engine-level limiter to set a cap just below the achievable average framerate as to prevent maxed GPU usage, keeping the render-queue clear of extra pre-rendered frames in the first place.

This is probably also partly why Reflex can affect 0.1 and 1% lows in some cases (dynamic limiter calculations/detection lagging slightly behind the more abrupt instances of frametime variances/spikes).

LLM "On" and Reflex have nothing in common beyond attempting to reduce/prevent additional render queue-level latency in GPU-bound scenarios.

[brownvim]

This Reflex bug happened with Arc Raiders after the Cold Snap update. Having Reflex plus Boost on made the game micro stutter and 1% lows a lot worse.

It was such a smooth running game before.

[Chief Blur Buster]

Your link to the Open Source Display article has a typo in it but I found and read it.

Thanks for the heads up, I have fixed it!

Lots of pros and cons to different methods and even games still. Looking forward to seeing a really bright 60 Hz mode on the Pulsar though — as I tried ShaderBeam and the brightness hit was too much for my liking, but it cleared up the 120 Hz artefacts in Nioh 3 compared to the Pulsar. Also not totally reliable with some noticeable flicker. Need to test it more.

It's definitely not as perfect-sync as hardware. Also, it works better if your game is low priority and ShaderBeam is high priority.

But, when it works well, it works fantastically and does things some screens cannot do. You can also do some adjustments to brightness by adjusting "Brightness vs Persistence" to a value closer to 1.0

I don't think the Pulsar monitor should be marketed for competitive shooters, its strengths are in other games. They did only demo it at CES using Anno with a lot of panning. However, all of NVIDIA’s marketing was really pushing the competitive shooter angle hard, which I think was quite overdone.

I can see how some people have been let down by it, but I also think it's also a mix of how different people see blur — whether it's during tracking moving objects or fast fixed camera panning / look-around movements.

I think some articles need to be created about this. I have a planned TestUFO inspired by discorz' image with multiple bands of different stationary/motions.

[mawi]

This Reflex bug happened with Arc Raiders after the Cold Snap update. Having Reflex plus Boost on made the game micro stutter and 1% lows a lot worse.

It was such a smooth running game before.

Its really weird. I tested a lot more last night and its not just Reflex which brings the 1% lows down. What I wrote yesterday is still the best setting for me although it gives me 13ms latency while with reflex I have 8ms (in comparable quiet scenes at spawn position). I tried to pin my framerate down 10 fps below max via nvcp and turn vsync off... et voila, same 8ms latency as with reflex (nv low latency still "on").

But why on earth... 1% low then drop down the exact same way as with reflex... I know reflex also pins down max fps 20-25 fps below 360. So these appear to be similar mechanisms.
But all I am trying to do is game without reflex and staying away from the boundaries where vsync becomes important while gsync does the rest on all the framerates below. Vsync is adding latency when its turned on even if you are below the max refresh rate so it would be nice to be able to stay away from this border without destroying the 1% lows.

CS internal frame cap is even worse btw.

I will stick with yesterdays settings as they still look and feel the best but I still think there is room for optimization.

Anyone any more ideas?

[shapaco]

Kyube, thanks for all the recommendations regarding backlight strobing monitors. I was honestly close to returning the Pulsar, but since yesterday I’m really glad I didn’t. Thanks to Mawi’s post, I finally saw what the Pulsar is all about.

I turned off Reflex + Boost in my games and changed the global Low Latency Mode in NVCP from “Ultra” to “On.” That improved my 1% lows significantly, and suddenly all my games started to look absolutely amazing — much better than during my initial testing.

Another thing I unfortunately hadn’t really considered was whether the game was running in borderless windowed or fullscreen mode. In The Finals, I switched to exclusive fullscreen and completely disabled Reflex. Only then did the game finally start looking the way it’s supposed to.
With Reflex set to Off, I get the best 1% lows, but at the cost of around 10–15 ms of additional latency. Setting Reflex to On seems to be a good compromise — slightly lower 1% lows, but noticeably better latency.

At times it genuinely feels like I have a huge CRT monitor in front of me. The improvement varies from game to game, but overall it looks fantastic (even though The Finals is still a bit of a problem child).

Thanks again, Mawi and Kyube, for your input

I do have another question:

I’m currently a bit addicted to a browser shooter called Deadshot (deadshot . io). I tried enabling Pulsar in this game, but since it runs in a browser, I can’t properly limit the framerate in NVCP or RivaTuner for G-Sync to work correctly.

I also downloaded the standalone client, which pushes the FPS up to around 1500, but that didn’t really improve things in ULMB2 mode either.

How would you approach this?

[mawi]

Thanks again, Mawi and Kyube, for your input

I do have another question:

I’m currently a bit addicted to a browser shooter called Deadshot (deadshot . io). I tried enabling Pulsar in this game, but since it runs in a browser, I can’t properly limit the framerate in NVCP or RivaTuner for G-Sync to work correctly.

I also downloaded the standalone client, which pushes the FPS up to around 1500, but that didn’t really improve things in ULMB2 mode either.

How would you approach this?

I am happy you have similar experience
Regarding Full screen exlusive... at least on CS2 i did not see a similar benefit. Its the same avg fps, same 1% lows and same latency if I use it or not. But sure, that can vary from game to game.

Regarding your browser game... maybe a stupid idea but: Have you tried Lossless Scaling? You could pin your browser fps to 360 with the variable frame generation (not 2x / 4x / 6x but applying a target fps). I would not use the gsync option in there.. It never worked for me.

[brownvim]

I don't think the Pulsar monitor should be marketed for competitive shooters, its strengths are in other games. They did only demo it at CES using Anno with a lot of panning. However, all of NVIDIA’s marketing was really pushing the competitive shooter angle hard, which I think was quite overdone.

I can see how some people have been let down by it, but I also think it's also a mix of how different people see blur — whether it's during tracking moving objects or fast fixed camera panning / look-around movements.

I think some articles need to be created about this. I have a planned TestUFO inspired by discorz' image with multiple bands of different stationary/motions.

Look forward to reading them, I like being able to track objects in game again.

I really like the technology (ShaderBeam and Pulsar) and interested to see how they evolve.

Happy the shooter fans are finding solutions too.

[liquidshadowfox]

After testing in hunt showdown, I see I get better motion clarity with reflex enabled at 120 fps than I do at 240 fps (with riva tuner cap using reflex markers). I don't know if it's because the gsync module is too sensitive to fps fluctuations or if @discorz's discovery about the pulse width seeming higher at higher fps has to do with it. When I go to the practice mode, I can clearly see the scratches on the dummies as I pan the camera back and forth with a 120 fps cap with reflex, the moment I go to 240 fps cap I see it much blurrier and looking at the FPS graph from the monitor OSD I see the fps is almost as stable as the 120 fps graph. I think they need to implement a toggle to allow the strobing to be a little more lax when it does the compensation pulse because at this rate many people will say it's not as good as reviewers make this strobing out to be (unless they deep dive into this forum and read the research everybody is putting in). They should make a pulsar aware version of reflex because in some games reflex works great (hunt showdown for example) and the fps isn't all over the place and then other games it absolutely ruins the frame pacing and you have to turn off reflex and rely on something like driver level low latency mode = on. Also keep in mind on top of all this, you have window security updates that are causing frame drops on nvidia GPUs (KB5077181) or Nvidia driver updates that ruin Vsync on + frame gen (anything past driver 591.67 which doesn't "support gsync pulsar" but I've been using it just fine). what a mess lol

[mawi]

After testing in hunt showdown, I see I get better motion clarity with reflex enabled at 120 fps than I do at 240 fps (with riva tuner cap using reflex markers).

Interesting. But could it be that your 1% low with reflex on were always around 120 fps and by capping the average fps to 120 you generate an overall smooth experience (similar to what I had on quake rtx)?
I think that might be a solution for any non-esport shooter game...why not play at stable 100-120 fps which looks perfectly clear on pulsar instead of having all these ups and downs.

But for esport titles its not just the visuals, many gamers feel the latency which are purely based on fps... even if a pulsar screen makes 110 fps look perfectly clear. They really need the calculated 240fps... and in this case, similar high 1%fps to make it look as it should.

I totally agree with you there should be a reflex option which is somehow aware of a pulsar display that optimizes whats possible while keeping the 1% fps at a level that leaves the "wow" effekt alive. Or... on the display side, whatever is possible there to deal better with 130fps lows when the average is at 250...

I still believe all the testers and probably nvidia theirselves did most of the tests with highend GPUs. The typical experience of an rtx 3060 gamer etc is not covered yet. Lowering the resolution could help but all four plusar displays are the same native wqhd display. Who voluntarily goes to FHD with such a panel...

[liquidshadowfox]

In hunt showdown my 1% lows are easily above 165 fps with 9800X3D + 5090 but the motion clarity is actually LESS as you go up in refresh rate. I tried 144 and the scratches seemed less defined than 120 fps and my riva showed my 1% lows were matching the 144 but my 0.1% lows weren't at 144 fps (I think they dipped as low as 98) so it could be either A) nvidia Gsync module is VERY picky about the frame times being close to the 0.1% lows otherwise it'll do the compensation pulse B) my nvidia driver that doesn't support pulsar is affecting the pulse timings at higher refresh rates (but no option to update without giving up Vsync + Gsync + reflex + framegen and in some games the Vsync + Gsync is required to get good frame pacing) C) This is a firmware bug they might address in the future

Either way, I'll probably play more with this as time goes on and see if I find any "gotchas" to getting this to work perfectly. I think nvidia is supposed to be releasing a new driver this friday so hopefully they will fix the Vsync + frame gen issue.