Most competive configuration for FPS Games (155hz)

[forii]

People wire an led into their mouse and use expensive high speed cameras to see how low a change occurs on screen.

I DID NOT CHANGE ANY AUDIO DRIVERS TO MSI. I tested my nvidia audio and it works. The card is made to support MSI, anything that isn't won't function at all.

Use RTSS as an fps limiter because NVCP is a driver level fps limit that will give you more fps lag than a cpu level limiter like RTSS. It's unfortunate the in-game limiter isn't a good option as they usually have the lowest input lag.

The difference between 144 and 140 is about 0.2ms. You could use RTSS fps limiter and get the extra 0.2ms without tearing if your fps wasn't 144 stable.

I am wrong. Vsync off is slightly better, by 1-3ms according to that page.
https://blurbusters.com/gsync/gsync101- ... ettings/9/
"So, for competitive players, V-SYNC OFF still reigns supreme in the input lag realm, especially if sustained framerates can exceed the refresh rate by 5x or more. However, while at higher refresh rates, visible tearing artifacts are all but eliminated at these ratios, it can instead manifest as microstutter, and thus, even at its best, V-SYNC OFF still can’t match the consistency of G-SYNC frame delivery."

Consistency is very important when it comes to aim. The current best option in my opinion would be RTSS's SSYNC so that you have no tearing without the input lag of vsync, but you will have the input lag of a cpu level fps limiter. It's the best option because the in-game fps limiter doesn't work. If it did work the best option would be to set it to 140 or even 120 and use gsync + vsync NVCP, in-game vsync disabled. The difference between 140 and 144 is only about 0.2ms.

I only recommend using fast sync if you're using RTSS's SSYNCx2 because a game's fps limiter isn't working or isn't good, but this requires you to have a stable FPS twice that of your refresh rate. which would be 288 stable on your 144hz screen.

You will have the lowest input lag with vsync/gsync disabled by a hair, but how it effects your aim isn't worth it.

edit: cpu based fps limiters add 1 frame of input lag so you're looking at ~7ms extra input lag for the perfect consistency.

Alright I have turned MSI to GPU (only) and restart my PC, dont feel any change for now, but will keep using/testing it.

if it comes to RTSS vs NVCP i heard NVCP gives you more stable framerate... But if RTSS is better so I will use it.
Now... the Scanline... how do I set it? It can't be run with any form of g-sync/v-sync
https://www.reddit.com/r/nvidia/comment ... is_a_game/

I found we need to use -30?? or how do I setup it?

"2a. FPS at the image refresh level -> set "Scanline sync" to a value between 1060-1120 for a 1080p monitor, for other resolutions -50."

I suppose to set scanlince sync at 1080?? or what? i dont get from where we got this huge number

Im kind of confused now...
So for now what I know (correct me If am wrong)
1) FPS limiter in game is the best for low input lag, if it doesnt work corecctly we should use RTSS
2) FPS Limit should be always bellow HZ of monitor (unless it can only 5x double refresh rate) ->> no matter we dont use VSYNC?????????????? (because then I could just dont use any form of g-sync/v-sync, use Ultra low latency + 0 fps in RTSS, so i could have even 200+ fps
3) Optimal option, a little bit bigger input lag but best visibility -> free sync (g sync compatible) + vsync in NVCP + null + RTSS cap at 140 fps.
4) Scanline is still a "?" to me

[forii]

Edit.

I tested again the FPS limiter in game, actually it works, I mean... When I set 140, it doesnt lock fps at 140 but it lock then max at 143, it doesnt go higher than that... I tried to do that with 138,137 etc, and it only works next with 125 fps lock...

Can i stay with that 143? If it doesnt go higher - because I saw my fps - then its perfect???
I remember what you said about fps going higher than refresh rate and pops out v-sync, but I would rather to have 143= 143hz than 125hz

So what I have tested now to be clear: G sync compatible + v sync in NVCP (no in game) + NULL (latency, at ultra) + fps limiter in game at 143 = so i have now lowest input lag? plus ofc MSI tool on GPU

[ELK]

RTSS's accuracy is in the microseconds, and has extremely good frame pacing. It does add 1 frame of input lag versus an in-game fps limiter but NVCP's will add 2.

Set SSYNC above your vertical resolution. You don't want to see a tear line on the bottom of your screen, but you want to keep it close, if you push it too far it the tear line will be at the top of your screen. If you're 1080p try 1090. If you see the tear line at the bottom of the screen increase it. If you increase it too much you will see the tear line at the top of the screen so decrease it.

-30 is used to pair with vsync. Chief suggests 10% of the screen at lower refresh rates or 20% at higher refresh rates. This greatly decreases vsync's input lag to 10% or 20% of a frame instead of the 2~3(yeah 200% 300%) it can do. I prefer for tearing to occur than my fps to be cut in half by vsync, but this will only occur if your fps isn't stable.

To explain it a bit better. Your graphic card communicates to your gpu like a film reel. The pictures aren't connected but there is a blank space inbetween them. You're pushing the tear line into the blank space. Setting a monitor running 1080p to 1060 will always result in a tear very near the bottom. setting it to 1120 will push the tear line past that blank space into the top on my monitor but some monitors you could set it to 1400 and not have a tear line at the top. It all depends on how big the vertical blank is, but ideally you want to keep the tear line closer to the bottom (for example 1090) which allows everything more time to render.

1) in-game fps limiter is almost always the best option but some game's have poorly built fps limiters. It's rare that's it's not the best option
2)There are many different setups and fps limit should only be below hertz when using gsync.
2a) I'm recommending a setup with the highest consistency with low input lag. Consistency is going to improve your "precision" not your accuracy. Accuracy comes down to how good you are with a mouse. Precision is going to come done to frame consistency. If you are a robot and can perfectly aim at the center of a head flawlessly every time with 100% accuracy, how tight the grouping of your shots are depends on the consistency of your monitor. Basically with perfect cosistency you have to aim within 30x30 hitbox for 100% headshots. With poor consistency you may have to aim within a 20x20 square of a 30x30 hitbox for 100% headshots.
3)This is the best option if you don't have stable fps.
4) again with the many different setups.

I'll post every setup and explain when and why you should use them. After I reply to your next post. I do this 1 step at a time XD

[ELK]

Edit.

I tested again the FPS limiter in game, actually it works, I mean... When I set 140, it doesnt lock fps at 140 but it lock then max at 143, it doesnt go higher than that... I tried to do that with 138,137 etc, and it only works next with 125 fps lock...

Can i stay with that 143? If it doesnt go higher - because I saw my fps - then its perfect???
I remember what you said about fps going higher than refresh rate and pops out v-sync, but I would rather to have 143= 143hz than 125hz

So what I have tested now to be clear: G sync compatible + v sync in NVCP (no in game) + NULL (latency, at ultra) + fps limiter in game at 143 = so i have now lowest input lag? plus ofc MSI tool on GPU

So 143 and 125 rings bells right off the bat. It's limiting to 1000/7 (142.857...fps) and 1000/8 (125.000fps). We have seen in-game fps limiters have +-2 variance, and we have seen some that have less than 1 variance meaning 143 work flawlessly. It could be reporting 143 or even 142 and actually be reaching or exceeding 144. Test equipment would be needed to test the input lag, and to see if it's reaching the vsync ceiling. If it does work and never reaches 144 thus the vsync then then 140 in-game limit with gsync enabled would be the absolute best option in the world. From what I've seen the 1000/x fps limiters usually work very well. So my hopes are up.

There are tests you can do yourself to suggest if it works or not. Enable gsync +vsync (NVCP only) and use the 140 fps limit in-game. Do you notice stuttering. If so that suggests it's flipping between gsync and the vsync ceiling. Is there a BIG difference in input lag with gsync enabled and the 140 fps limit vs gsync disabled. This would suggest gsync is stuck at the vsync ceiling. Is there a BIG difference at 125 than 143 with gsync enabled? This would also suggest that gsync is stuck at the vsync ceiling. Were talking maybe 20ms difference. It's a BIG difference and HIGHLY visible to competitive gamers. If the difference is "I can't really tell" then it's probably always staying in the gsync range.

[forii]

RTSS's accuracy is in the microseconds, and has extremely good frame pacing. It does add 1 frame of input lag versus an in-game fps limiter but NVCP's will add 2.

Set SSYNC above your vertical resolution. You don't want to see a tear line on the bottom of your screen, but you want to keep it close, if you push it too far it the tear line will be at the top of your screen. If you're 1080p try 1090. If you see the tear line at the bottom of the screen increase it. If you increase it too much you will see the tear line at the top of the screen so decrease it.

-30 is used to pair with vsync. Chief suggests 10% of the screen at lower refresh rates or 20% at higher refresh rates. This greatly decreases vsync's input lag to 10% or 20% of a frame instead of the 2~3(yeah 200% 300%) it can do. I prefer for tearing to occur than my fps to be cut in half by vsync, but this will only occur if your fps isn't stable.

To explain it a bit better. Your graphic card communicates to your gpu like a film reel. The pictures aren't connected but there is a blank space inbetween them. You're pushing the tear line into the blank space. Setting a monitor running 1080p to 1060 will always result in a tear very near the bottom. setting it to 1120 will push the tear line past that blank space into the top on my monitor but some monitors you could set it to 1400 and not have a tear line at the top. It all depends on how big the vertical blank is, but ideally you want to keep the tear line closer to the bottom (for example 1090) which allows everything more time to render.

1) in-game fps limiter is almost always the best option but some game's have poorly built fps limiters. It's rare that's it's not the best option
2)There are many different setups and fps limit should only be below hertz when using gsync.
2a) I'm recommending a setup with the highest consistency with low input lag. Consistency is going to improve your "precision" not your accuracy. Accuracy comes down to how good you are with a mouse. Precision is going to come done to frame consistency. If you are a robot and can perfectly aim at the center of a head flawlessly every time with 100% accuracy, how tight the grouping of your shots are depends on the consistency of your monitor. Basically with perfect cosistency you have to aim within 30x30 hitbox for 100% headshots. With poor consistency you may have to aim within a 20x20 square of a 30x30 hitbox for 100% headshots.
3)This is the best option if you don't have stable fps.
4) again with the many different setups.

I'll post every setup and explain when and why you should use them. After I reply to your next post. I do this 1 step at a time XD

Alright, so I still got questions...
1) Do I need to use g sync with or without v-sync in NVCP + NULL at ultra with that Scanline in RTSS? (1090 for example, or more).
And do I need to cap fps in RTSS bellow hz or what do I do with that 1090 scanline? just set the value and test if all is fine?
And with using scanline I will have 0 tear, zero input lag and thats the best method for people who dont have g sync?
2) What about stable fps vs scanline? Because you said to my question "3) Optimal option, a little bit bigger input lag but best visibility -> free sync (g sync compatible) + vsync in NVCP + null + RTSS cap at 140 fps."

3)This is the best option if you don't have stable fps.

3) So if I would have stable fps above 144 (it needs to be always?), and they would NEVER go down bellow 144hz I do not need to use G sync with V sync as we mention above as a best consistency setup with low input lag? Is that correct?
So then the v-sync off would be best for input lag + frame limiter in game, how it goes vs scanline?

4) If we have not stable fps - that means - our fps drops sometimes bellow refresh of monitor (for example 135 fps drop) then its almost always must have to use g-sync, unless we dont care about stutter and tearing?

5) I saw v-sync adds input lag, but g sync doesnt, why we use v-sync with g sync, it v-sync only is used if fps goes above 144. I dont get it.
Can't I use only g sync + fps limiter in game? + ofc NULL at ultra (which I will use for now in every situation)

6) If it comes to test If 143 cap in game is enough I think the best would be to make it higher, for example I can do 160 fps cap and then I can test if this is slower compare to 143 fps cap (or same - if same, then 143fps cap exceed refresh rate and v sync is turning on) - then I would use rtss then

[forii]

Alright so fps limiter in game is bugged, fps exceed the refresh rate anyway (i saw when I turned the fps statistic in game) - but didn't see their exceed on RTSS info, weird..

I am not sure If i understand correctly, thats why I will wait for the answer. But for now Im using only 1090 scanline in RTSS (framerame at 0) + NULL at ultra + no g-sync or v sync anywhere.

I dont see any difference actually compare to g sync, cant meassure input lag.

Also I forgot to mention that I use Intelligent standby list cleaner ISLC.exe

I hope this app does work with MSI and NULL because it does some custom time resolution which I can put off

[ELK]

Alright so fps limiter in game is bugged, fps exceed the refresh rate anyway (i saw when I turned the fps statistic in game) - but didn't see their exceed on RTSS info, weird..

I am not sure If i understand correctly, thats why I will wait for the answer. But for now Im using only 1090 scanline in RTSS (framerame at 0) + NULL at ultra + no g-sync or v sync anywhere.

I dont see any difference actually compare to g sync, cant meassure input lag.

Also I forgot to mention that I use Intelligent standby list cleaner ISLC.exe

I hope this app does work with MSI and NULL because it does some custom time resolution which I can put off

If you use the 125fps limit does it ever reach 144fps?

You can't feel any difference between 143fps limit with gsync/vsync on vs gsync/vsync off?

Setting it to 0.5ms will just increase cpu load. It is extremely rare for a game to not set it to 1ms by default. Games aren't programmed to utilize 0.5ms and if they are it will automatically set it to 0.5.

Everything will work while your video card is in MSI mode except virtual machines. Everything will work with NULL.

I may not be right but from what I know standby memory is a combination of superfetched, prefetched, and closed programs. Superfetch and prefetch load commonly used stuff into your ram so that it will open quickly if you decide to open it but this puts extra work on your hard drive which will decrease performance while gaming. You can disable these in the registry. When you close a program instead of changing the ram to all 0s it just leaves it there in standby meaning anything can overwrite that portion of ram if it wants more ram. Maybe there is something I don't know though.

[ELK]

I'm leaving non VRR monitors out of this XD. Vsync should be enabled in the NVCP and disabled in-game for options that have gsync/vsync enabled.

gsync/vsync disabled, NULL, in-game fps cap to keep gpu under 90%.
+Lowest possible input lag
-Tearing and poor consistency
-Can look stuttery
!This is the best option where accuracy isn't required for example a competitive real time strategy game

gsync/vsync enabled, NULL, in-game fps cap 3 below refresh rate and keep gpu under 90%
+2nd lowest possible input lag
+No tearing and high consistency
!This is the best option where accuracy is important for example first person shooters.

gsync/vsync enabled, NULL, in-game fps cap below refresh rate and keep gpu under 90%
+3rd lowest possible input lag depending on in-game fps limit options*
+no tearing and high consistency
!This is the 2nd best option for shooters depending on the in-game fps limit options*
*RTSS will add one frame of input lag. This is 240(~4.2ms) 155(~6.5ms) 144(~7.0ms) 120(~8.3ms). Using an in-game fps cap of 120 on a 144 monitor will result in +~1.4ms of input lag while RTSS at 141 will result in +~7.1ms.

gsync/vsync disabled, NULL, RTSS SSYNC and keep gpu under 90%
+4th lowest possible input lag
+high consistency
+-no tearing if frame rate is stable
--only ~0.1ms better than the option below
!These are the best options if there is not an in-game fps limiter

gsync/vsync enabled, NULL, RTSS fps cap 3 below refresh rate and keep gpu under 90%
+4th lowest possible input lag
+no tearing and high consistency
!These are the best options if there is not an in-game fps limiter

So in the case of modern warfare, being a FPS, you're best off using the in-game fps limit of 125 with gsync/vsync, NULL, as long as your gpu is staying below 90%. This will result in about +1.1ms of input lag. If the in-game fps limiter doesn't work at 125 (stays below 144) than the best option is RTSS at 141 with gsync/vsync, NULL, as long as your gpu is staying below 90%. This will result in about +7.1ms of input lag.

[forii]

I'm leaving non VRR monitors out of this XD. Vsync should be enabled in the NVCP and disabled in-game for options that have gsync/vsync enabled.

gsync/vsync disabled, NULL, in-game fps cap to keep gpu under 90%.
+Lowest possible input lag
-Tearing and poor consistency
-Can look stuttery
!This is the best option where accuracy isn't required for example a competitive real time strategy game

gsync/vsync enabled, NULL, in-game fps cap 3 below refresh rate and keep gpu under 90%
+2nd lowest possible input lag
+No tearing and high consistency
!This is the best option where accuracy is important for example first person shooters.

gsync/vsync enabled, NULL, in-game fps cap below refresh rate and keep gpu under 90%
+3rd lowest possible input lag depending on in-game fps limit options*
+no tearing and high consistency
!This is the 2nd best option for shooters depending on the in-game fps limit options*
*RTSS will add one frame of input lag. This is 240(~4.2ms) 155(~6.5ms) 144(~7.0ms) 120(~8.3ms). Using an in-game fps cap of 120 on a 144 monitor will result in +~1.4ms of input lag while RTSS at 141 will result in +~7.1ms.

gsync/vsync disabled, NULL, RTSS SSYNC and keep gpu under 90%
+4th lowest possible input lag
+high consistency
+-no tearing if frame rate is stable
--only ~0.1ms better than the option below
!These are the best options if there is not an in-game fps limiter

gsync/vsync enabled, NULL, RTSS fps cap 3 below refresh rate and keep gpu under 90%
+4th lowest possible input lag
+no tearing and high consistency
!These are the best options if there is not an in-game fps limiter

So in the case of modern warfare, being a FPS, you're best off using the in-game fps limit of 125 with gsync/vsync, NULL, as long as your gpu is staying below 90%. This will result in about +1.1ms of input lag. If the in-game fps limiter doesn't work at 125 (stays below 144) than the best option is RTSS at 141 with gsync/vsync, NULL, as long as your gpu is staying below 90%. This will result in about +7.1ms of input lag.

Damn man!! This is WHAT my MIND wanted to read!!!!!!!

It is hard to me to end the topic, so I think i got last question... it is about the fps limiter in game:

I checked the fps limit in game and It seems that the game is blocking my FPS at 143 not because of the fps limiter inside, it's probably due to v-sync along with g-sync or maybe of the NULL, It blocking my fps itself (without fps limiter), so it cant go higher than that, that is some weird magic stuff I dont get (yet).

And also my smart friend told me in other language this (I will translate it):

If you have V-Sync turned on and fps does not break through the monitor refresh, then there is no point in playing additional blocking frames, for 144 / 165Hz if it holds 143-144fps / 164-165fps it is very good, even if 145 / appears for a second 166 does not mean that in fact the number of frames has exceeded the range, the software measurement frametime and framerate has never been perfectly accurate.

^ in my case actually the fps doesnt go from 143 to 164-165, I can only choose 125 cap/ 143 cap or 165 cap (dunno why but it seems that 143 is fine, and will be more smoother/faster than 125 fps cap (125hz)

+ His second quote about 143fps cap which is near 144hz

There is no such need, as I said while it is within the range there is no problem, once limiting FPS to reduce the range of VRR referred to the classic G-Sync, which had kicked buffering which was manifested by flickering and clipping, after updating Windows 10 1803 and WDDM to 2.4 G-Sync flickering has been reduced, and clipping has been eliminated, in addition using G-Sync Compatible based on Freesync flickering you should not experience at all.

(i hope google translated it correctly and you understand)


+ his quote about NULL at Ultra vs ON

Low Latency mode is not always beneficial in Ultra mode, it is best to switch to On then it gives you frame buffering at the level of 1 piece.

^ which I dont understand but I think NULL at On for me will be still fine after the MSI + a lot of tweaks in my system (Intelligent standby list cleaner ISLC)

[ELK]

Sounds like gsync is reaching the vsync ceiling then and you should the 125 cap.

"If you have V-Sync turned on and fps does not break through the monitor refresh, then there is no point in playing additional blocking frames, for 144 / 165Hz if it holds 143-144fps / 164-165fps it is very good, even if 145 / appears for a second 166 does not mean that in fact the number of frames has exceeded the range, the software measurement frametime and framerate has never been perfectly accurate."

You want to block a few frames to avoid the vsync ceiling while using gsync. The vsync ceiling will frame(s) of input lag. This is why 141 is better than 144 when using gsync. Yes it would be ideal to have an in-game fps limit of 141 vs 125 BUT the game does not offer 141. Using RTSS to get 141 instead of 125 would result in higher input lag due to the fact that RTSS will add 1 frame of input. Another big reason is that when your gpu is bottlenecking you will have much more input lag than if it were running under 90%. Nvidia has even commented on this and says it is expected.

"There is no such need, as I said while it is within the range there is no problem, once limiting FPS to reduce the range of VRR referred to the classic G-Sync, which had kicked buffering which was manifested by flickering and clipping, after updating Windows 10 1803 and WDDM to 2.4 G-Sync flickering has been reduced, and clipping has been eliminated, in addition using G-Sync Compatible based on Freesync flickering you should not experience at all."

Translation maybe isn't the best. Sounds like people were limiting the fps to keep it in range because if they didn't it would flicker? I know the flickering problem wasn't effecting every monitor. We're not limiting fps to prevent flicker. We're limiting fps to avoid the vsync ceiling which adds frame(s) of input lag.

"Low Latency mode is not always beneficial in Ultra mode, it is best to switch to On then it gives you frame buffering at the level of 1 piece."

Nvidia has cofirmed that it will not benefit every game. It depends on the game's engine I believe. I would leave it on ultra. I would assume that the games it does not benefit setting it to 1 also won't have any benefit but I could be wrong....

Your gpu needs the data from the cpu before it can start drawing.
When NULL is set to on your cpu can start processing data for the next frame for your gpu as soon as your gpu accepts the previous cpu data. For example, the gpu recieves the cpu data for frame 100 then the cpu instantly starts working on frame 101.
When NULL is set to ultra your cpu can start processing data for the next frame as soon as your gpu wants to begin the next frame. For example, the gpu recieves the cpu data for frame 100 the cpu idles until the gpu wants data for frame 101 and process it then.