Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by 321Boom, Jun 25, 2017.
Yes, for the hundredth time.
RTSS is superior in every way compared to NV inspector.
To me, this sounds like a smoker who's being told to hang an air freshener on the mirror, but refuses to do so because the air freshener adds weight to the car and thus will increase fuel consumption and lower car acceleration.
The claim that closing RTSS somehow lowers latency on the mouse is like saying that removing the air freshener makes the car accelerate faster.
We don't know if RTSS is polling for processes. Even if it did, that doesn't affect anything. Lots of processes and drivers in Windows are polling for random things all the time. It does not affect anything.
But is RTSS even polling at all? Why would it? As said previously, Windows can tell you when a process is starting. You register a callback, and Windows will call it, telling you which process is starting. There is no polling required.
All of this doesn't take anything away from nvidia's driver-level limiter, btw. It's a perfectly fine user preference, everyone can use their computer however they please. People have been using profile inspector to enable the limiter for a long time now, and they're happy with it. Again, that's perfectly fine and no one has the right to force anyone to use something they don't want. However, misinformation is not a good thing. If you say that RTSS is increasing mouse latency and lowers performance, that claim needs to be backed up. Either with data, or if that's not possible, then with some repeatable test that people can perform.
RTSS might in fact lower game performance. I don't know. But I would like some test to go along with that claim. Like a benchmark or something.
Hah, that reminds me of an episode in That 70s show where kelso removed car's seatbelts to make it go faster.
My intention has not been to demean or "one up" Mda400.
I truly have wanted to communicate. I just wish he would pay more attention to the results we have been discussing.
I post results and references (multiple times), and the contents of his replies make it seem like I've never offered them.
Mda400, your communication style is a bit perplexing; it's like you're never fully paying attention, let alone bothering to study the sources we are referencing. Every reply is like a reset back to the initial conversation.
I feel like I'm talking in circles at this point...
I hope not. I'm trying to keep myself as formal about this as possible.
This isn't communicating? You're saying you want me to take your results at face value when I know there's other variables that aren't detailed in your articles. "Good enough" is a very opinionated statement.
In other words, let's agree to disagree. We are discussing two sides and I choose to advocate the inspector side while you advocate the RTSS side. Yes you have colorful graphs and I only have words, but that doesn't bar me from putting in my opinions. You want me to eat crow, video is the only thing that is believable for something this specific and caused by many factors.
That's ironic, considering we have gone into such detail about what causes input delay that anything like a "car freshener" would skew the results.
While the application is hooked to another process, it needs to poll that process and the hardware its using for its state. RTSS may take very little cycles to limit the framerate and poll the application, but it is doing so nonetheless.
If it was perfectly fine, then we wouldn't be trying to prove which one is better. People don't use a lesser product for the same cost (in this case, free). Until that end is why we debate..
Finally something we agree on. I have given these and so have you. I need to make some colored graphs though.
I did say that was on old hardware and it would not be very noticeable with today's hardware. Like running a 1000hz mouse on an old 1.6ghz pentium M versus an i7-3770k.
I tried Mda400, but since I know you aren't actually going to read through or acknowledge any of my further clarifications or explanations, and you haven't bothered to assimilate my previous points (going so far as to completely misquote my article), I'm checking out of this conversation now; it has gone on long enough.
Hopefully other users will be able to draw their own conclusions on this subject from our posts thus far; there is plenty of material.
Which is what I basically have said in the end to everybody I have debated these two solutions with. There are many variables to reducing latency, some you may not use that I use otherwise and vice versa.
If you set a latency floor you are trying to achieve with a certain group of settings then it may be easier to debate which is the more superior solution, but latency is a major component of computing and with the claim of settings depending on other settings (fast sync + v1 limiter for example), then we don't know what other settings might depend on other settings.
I still discover settings that affect delay that I never realized were there before...
That's just insulting. You don't even realize the tremendous amount of work that went into producing those graphs. How much time it took, and the amount of data it required.
You're insinuating that all that was done is "draw some colored graphs".
You have NO idea.
No, that is not why we debate. We debate the issue of having provided proof, and you claiming it's not proof, and furthermore downplaying the proof as "colored graphs", while having no clue what it took to produce them.
Just ignore this dude, let him live in his own little world.
He says all the work that jorimt has done is just pretty coloured graphs.
He also said battlenonsense was making up **** with his graphs too.
Yet he makes all these claims about RTSS but does not even provide proof.
He is hopelessly dense.
@jorimt sorry you have to deal with people like this, I guess that's life though.
Since this thread was already answered, lets take our argument to PM's if you have a problem with me.
If the drivers limiter is so good as you claim why isn't NVidia including the limiter in NVCP?
I did switch between the drivers and RTSS back in time, input felt always more responsive with RTSS.
Using neither now though, GSync+VSync and sometimes ingame limiter.
I've done countless tests with framelimiters and RTSS is always the better option in every game.
The only problem I have with framelimiters is I can't use Adaptive Vsync because it always detect framerate is under refreshrate and tears even when framerate is locked 60 fps.
@RealNC & @jorimt Can you guys make a comparison between Fast Sync & Enhanced Sync in the future? Thanks.
Oh, well that's brand new. We'll have to wait to see specifics, but if Enhanced Sync is the equivalent to Fast Sync, both are just another name for true triple buffer V-SYNC.
I unfortunately don't have access to an AMD setup (I'm running a single Intel/Nvidia system), so I couldn't test it directly myself, but if it is what I think it is, the results would be virtually identical to my existing Fast Sync results/explanations in my G-SYNC 101 article.
I quick update to Enhanced Sync: it looks like it's a Frankenstein's Monster of Nvidia's Adaptive V-SYNC and Fast Sync according to this source:
It acts like Fast Sync with framerates above the refresh rate (avoiding locks to half refresh/missed delivery deadlines), but disables itself when framerates are below the refresh rate (V-SYNC OFF tearing).
Not sure how or if it is avoiding the microstutter that occurs with that method above the refresh rate (likely it doesn't and is identical to Fast Sync; syncing can only work so many ways without entirely new display hardware).
Interesting & thanks for sharing. Cheers.
Quick update @khanmein,
I recently spoke to the Chief Blur Buster, and we do indeed plan on testing Enhanced Sync in the future. No ETA.
I made a post about Enhanced Sync on our home page.
Woah, this is quite the definitive discussion on pros and cons of Fast Sync, V-sync, and frame limiters. My problem is more with killing both screen tear and microstuttering for a truly smooth experience with a regular monitor. I'm not a big fan of V-sync and Adaptive sync since it either doesn't always work right or still has tear. Fast Sync seems promising, but without a frame limiter games can run out of control and push the GPU way harder than it needs to on a 60hz monitor...Is there any combination of Fast Sync with a frame limiter close to 60fps (either RTSS or NV Inspector V1 or V2 or something else) that almost assuredly eliminates both while introducing as little latency as possible? It sounds like it might be too much to ask for a global solution that works for all games. What are your findings on this? I'm happy to sacrifice an extra frame of input lag to get rid of microstutter...that just bugs the hell out of me and kills my immersion, which is the whole point I play games in the first place. Thanks ahead of time!