Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by zandathrax, Nov 15, 2017.
Working fine here, no bugs yet.
I only need driver+physx.
Am I ok to go like this :
Do I need something more ?
Should I delete even the NVI2 folder ?
These drivers appear to be a vast improvement on the previous v388.13 (and v388.10) ones that suffering from one-second microstuttering (due to a Power polling bug) and constant microstuttering overall in many games.
Hollow Knight is a good test game for stuttering since it is a 2D side-scroller. That game runs smoothly on this driver with MSI Afterburner v4.4.0/RTSS v7.0.0 running whereas it would stutter every second on the previous two drivers, including the Hotfix. I was stuck on v388.00 WHQL for that reason.
I also tested Wolfenstein II which I had been playing fine with no issue using the v388.00 driver at 4K @ 2560x1440 DSR maxed out settings but with this driver the performance was shockingly poor. The framerate hovered around 55-60 fps but movement was extremely jerky and this was very obvious when standing still and panning the camera. This issue did NOT occur on v388.00! I was just about to reinstall v388.00 again when I decided to try playing the game at 2560x1440 and, lo and behold, with a much higher framerate of 90-130 fps the game runs smoothly and plays great. Not quite sure what is going on there unless DSR itself is bugged in this driver... I may need to try more 4K games.
Finally, I played about an hour of Assassin's Creed: Origins on the same 2560x1440 maxed out settings as I used with the v388.00 driver. This game also ran fine, pretty much exactly the same as the previous driver I was using so that is good news. I've decided to keep the drivers installed for now and try them with more games to see how they compare. The issue with Wolfenstein II at 4K is concerning though as that is the most drastic difference between this driver and the v388.00 one I was using prior.
P.S. I am using G-SYNC on a 165 Hz display and G-SYNC is showing as enabled in all three of the games I mentioned. G-SYNC is enabled for fullscreen and windowed applications with the NVIDIA global v-sync setting enabled and v-sync disabled in the game itself.
I had a lockup with these drivers in COD WW2. went back to the 385.69s..... all good again.
Is this at 4K because I had *exactly* the same issue until I lowered the resolution to the native one of my 27" ASUS ROG Swift PG279Q G-SYNC display. At 2560x1440 on maxed out settings the game plays fine but at 4K it's an atrocious jerky mess, which is obvious moving around but extremely noticeable if you stand still and slowly pan the camera around.
What is going on there?
driver is not bad. Pretty good.
After installing new drivers I started a new game in Wolf2 and played for a few hours and haven't had a single stutter. I did get a BSOD while playing Forza Horizon 3 though. Not sure what to make of it, never saw BSOD like this before.
I was playing at 4K simply because I had the headroom to run at this res and still have max settings. It's just a ego thing really, makes me feel good about having a GTX 1080 Ti and playing games at 4K (something the consoles struggle to do natively unless it's at 30 fps). I'm not sure if this driver was running slower than the v388.00 ones though because I had previously been getting over 60 fps (up to the *spoiler*Venus*spoiler* level; if I wasn't then I would have dropped the resolution down to 2560x1440).
In my experience, as long as the framerate stays over 60 fps then with G-SYNC the game runs extremely smoothly combined with the 165 Hz refresh. Even when it dips below 60 fps G-SYNC still helps the game stay smooth normally although I can feel the difference in the controls (using an Xbox One Elite controller by the way; no comments please as this is a single player game!).
However, with these drivers Wolfenstein II at 4K DSR feels horrible where with v388.00 the game ran as smoothly (albeit at lower framerates) as it does now at 2560x1440 using the v388.31 driver. Why this happens, I don't know, which I why I said that I may have to test other 4K DSR games to see if there is an issue with those on this driver.
4k DSR on 1440p display can look worse than 2560x1440 native resolution.
Working fine for me, are you using full screen?
Our testing confirms that Nvidia's 388.31 drivers boost Destiny 2 performance by over 30 percent
You didn't even open them, did you?
Could be because the trail exe is different to the launch exe?
Did not checked all the reactions here yet from page 1 till this page, but is the micro stuttering fixed even without msi afterburner ?
Speaking of driver regression, yes there were few moments, but nv fixed it. It wouldn't if we didnt complain..
Here it shows there was no real regression as overall, like dr_rus said..
Btw about 780gtx &farcry4, there was a bug i found a fix back then, reset ingame gfx cfg to high, then back to your setting, viola, fps back to normal and no stutter.
Driver is working pretty good in all my desktop systems. The only problem remaining is on my laptop, which is more Windows 10 FCU related issue. Forcing me to tick “Disable fullscreen optimization” on each game .exe to make games run smoothly…
This test is flawed too though. There's a number of games in it which weren't even available at the time of 780Ti release. What they do there is compare apples to oranges - you can't just change the benchmarking suit and say that the result is a sign of driver improvements / regression. And this in fact is what most such theories are based on: people are comparing older benchmarks with newer ones, and older cards are worse in these as expected. It's like running BF3 and then running SWBF2 on some Fury X and saying that the change is because of drivers - i.e. absolutely stupid.
Well, yes, I do agree with that as the downsampling can adversely affect the quality of the on-screen HUD and, of course, any text. This is why I prefer games that have built-in resolution scalers (like Assassin's Creed: Origins or Forza Motrosport 7 which go up to 200% of the native resolution) as you can have a sharper downsampled image and still return the same crisp HUD and text of the native resolution, which in my case is 2560x1440.
To be honest, I do tend to use 4K more for games that have poor AA support or shimmering issues, as downsampling helps reduce them, and mostly for older, less demanding games. Most new games use temporal AA which does a better job at removing jaggies (at the expense of more blurred textures) but the downsampling does help to compensate for that.
What was the reason for that recommendation again? Remember reading about it somewhere, and wondered if it could help issues with general Win 10 desktop systems, specially SLI and DSR...
I am the only one with a G-SYNC display that experiences odd performance issues in Unity Engine games on Windows 10 Pro v1709 build 16299.xx?
Specifically, the issue is that many games, such as the recent Outcast: Second Contact remake, can run at 165 fps one second then suddenly drop to 30 or even 24 fps, say, during a cutscene before returning back to 165 fps over the next minute or so. It's a really annoying issue that I've only been able to fully fix by forcing the exclusive fullscreen mode in affected Unity engine games. I've had to do this (so far) for Road Redempion, Dungeon 3 and several other recent Unity games. Not all games are affected though as Ori and the Blind Forest, Oddworld: New 'n' Tasty and The Last Tinker: City of Colors all work fine.
I'm not sure what the issue is, whether it's related to the way Windows 10 now treats fullscreen games (basically forcing them to be borderless fullscreen for Game bar compatibility) but it seems an odd coincidence that Unity engine games, by default, use borderless fullscreen mode rather than exclusive fullscreen like every other engine out there. I do not have this issue with games that use Unreal Engine (3 and 4), CryEngine, Frostbite or idTech for example.
I did see someone mention on the GeForce.com forum that G-SYNC isn't working properly in the 388.xx drivers and that it should only be enabled for fullscreen games not windowed ones (which presumably would include the borderless fullscreen ones that Unity uses?).
I have a theory regarding why the micro stuttering is gone in games like Hollow Knight or Rocket League - which could be due to them changing the behavior of "Optimal" Power setting meaning the card doesn't downclock as much as it used to.
I have not played Hollow Knight, Rocket League or other less demanding games but when I did I remember the clocks going really low since such games require very little of a 1080Ti. The GPU load would be as low as 15-25% even at 4K and the clocks would drop below 1300MHz due to this. What I did was enable Max Performance for said games which completely eliminated the stuttering I had.
Unfortunately this theory only applies to low demanding games (CS:GO would be another example).