Silk Smooth parameter in Nvidia drivers

Just did the NVProfile inspector and registry tweak and in my NVCP i noticed a new option with the following values. See the pic below! Drivers 418.81 WHQL. I think medium is the best option and btw it was the default value from the beginning when i 1st opened the NVCP. Also i cannot even change this setting in any program specific profile so.. i think this option is predefined and "locked" by nvidia so we can only view it in the Global settings section.

​

 

Does that option even work? I know some people say "I applied X or Y fix, now my games are super smooth!!!". To be honest, these "snake oil" fixes never worked for me and never made me see a change in games. I am not saying they don't or that OP's nVidia config didn't do anything, but I've tried this "silk smoothnes" before and never seemed to do anything for me. I'd like an option on the nVidia control panel to lock the frame times in a similar manner RivaTuner does to be honest. And no, the nVidia Inspector frame rate limiter is terrible and a buggy mess.

 

It all depends what you first select in nvinspector

Medium might be due to you selecting higher in nvinspector, happened to me too when I first enabled it with 417.22 driver.

silk parameter
0= enabled - low
1= medium
2= high

Also for it to work per profile, you need to manually enable in nvinspector game profile then it will show up in that said profile in nvcpl. -- The same way as you enabled it in global profile.
First that number level bellow and then silk parameter.

 

Interesting, might give it a go.

 

Btw this silk option is a NO GO for competitive FPS shooters like R6! Disabling this feature is highly recommended!

 

I finally tested it in a couple games. Makes zero difference. Off, low, ultra, no difference, neither in input lag nor frame pacing.

 

I recommend setting it to Medium or 0x00000001. Ultra or 0x00000003 cause some stuttering.

 

Pure smooth, anybody?

 

As I expected. Also, that link you posted of the nVidia guy saying "that is not there" or something along those lines is like saying "that does not work". If it worked, they would not have cut it from the nVidia control panel.

 

It wouldn't be the first time functional options are cut out. FPS cap for example.

 

Looks like a bunch of placebo to me.

 

nVidia would have given word already anyway if it worked at all.

 

It is kinda nice to see that people bash NVIDIA for this deeply hidden setting - it proves that some things in us humans do not change.

 

Without the gurus on this forum Nvidia wouldn't even know what theire settiings do! XD

 

nVidia does not know what it does? It does not even work lol xD.

 

It works. It’s not placebo and is the only way to achieve smooth framerate in Mafia 3.

 

Maybe it wasn't the Silk Smooth parameter that helped you but other settings you used. I had a similar issue with Ghost Recon: Wildlands where the game would stutter horribly often. Borderless+no v-sync+nVidia frame limiter V2 to 60 fps solve the issue completely. Now the game would not go to hell each time it would drop a frame or two. The thing that solved the issue was the borderless on Windows 10 forces Triple Buffer for games that don't have it and focing 60 fps on the nVIdia Inspector fixed the game. With RivaTuner it would get screen tearing, with then nVidia limiter it would not. That's the only case where the frame rate limiter in Inspector was useful because it never does a very good job.

 

I’m saying that is silk smooth and I’m 100% sure of that. Why are you arguing with me?

 

I am not "arguing". I am just giving my opinion. Why are you defensive about it? If you so sure that it works, why does it bother you that I question it? You think it works, other thing is that it really does. Placebo is a powerful thing, has been reported that it does not work. I've tried it and it does NOTHING related to frametime or input lag, the games run the same with or withou it. nViia Inspector shows all hidden GPU options, it does not mean that all of them work. Some games ignore some options entirely, such as V-Sync on some Ubisoft games.

Oh well, I better stop replying before you get mad.

 

Only way to know for sure is to record the output of the game with and without it. (Framerate plotted over graph, frametime plotted over graph) in a directly repeatable scenario that has little to no changes from run to run.

Though sometimes microstuttering exists even when it doesn't show up in the numbers. (Duplicate frames for example. Like in some of Yakuza zero's cutscenes)