NVIDIA discussion [2025]

[Nappe1]

Mine Asus VP220 isn't top noch either. Full hd 60hz without any variable refresh rate support is as bland as it gets.

Same problem happened also with HP x27q which is completely different beast. Also there was no difference between displayport nor hdmi cables and I tested all 4 display connectors in card.

Ddu in safe mode and reinstall 566.36 fixed it at once.

 

[Albuquerque]

Hmm...

Many months ago, thinking at least six and probably closer to nine or 12 months ago, my Monoprice 35" VRR 100Hz 3440x1440 DP-connected monitor starting having blackscreen issues every once in a while. What I eventually discovered after fighting it for a while was VRR support was somehow b0rked. I disabled VRR support in the driver and I've had literally no issues at all with my monitor ever since, and I'm 100% confident this started in the 560-series drivers. What's interesting is, at least for me, I've been keeping up the 570-seriies driver updates and have yet to (re-) encounter any blackscreen issues.

I have absolutely hit a few strange instabilities with my long-standing stable overclocking profiles in the 570-series drivers. My 2700MHz @ 950mV setting has been stable ever since I bought my 4090, yet somehow was no longer capable of starting a Cyberpunk 2077 saved game without crashing to desktop. I eventually re-settled on 2700MHz @ 975mV which is fine, yet is somehow measurably different from where I started.

 

[Man from Atlantis]

I have absolutely hit a few strange instabilities with my long-standing stable overclocking profiles in the 570-series drivers. My 2700MHz @ 950mV setting has been stable ever since I bought my 4090, yet somehow was no longer capable of starting a Cyberpunk 2077 saved game without crashing to desktop. I eventually re-settled on 2700MHz @ 975mV which is fine, yet is somehow measurably different from where I started.

Fortunately, I have no issues with the latest drivers. Last time I played CP2077 it was 2020 and had a 3080. I've been playing for around 50hrs so far, with the new OLED screen and PT update. I have the same with my long time stable OC, isn't stable in this game anymore. It was fine at Q2, Portal RTX ports. Have to revert to 2685MHz@950mV.

 

[Florin]

I have single monitor. Yet I still got the Black screen bug.

I've seen black screen behavior twice now on 5090. Not sure if this is the actual same bug because it seems more frequent for other people.

Turns out it wasn't a huge issue for me either as the Chrome Remote Desktop host service running in background remained accessible, so I just remote connect from phone or w/e. I alt-tab out of full screen game, then change desktop display res or refresh rate or something, and the black screen goes away. I can then revert to original display setting. All programs still running and stuff.

This is by no means an excuse for driver issues happening in the first place of course. It did reduce impact to a mere inconvenience for me.

 

[T2098]

I have absolutely hit a few strange instabilities with my long-standing stable overclocking profiles in the 570-series drivers. My 2700MHz @ 950mV setting has been stable ever since I bought my 4090, yet somehow was no longer capable of starting a Cyberpunk 2077 saved game without crashing to desktop. I eventually re-settled on 2700MHz @ 975mV which is fine, yet is somehow measurably different from where I started.

That's not entirely unexpected, as pretty much all silicon degrades over time due to age / electromigration. Most video cards I've ever owned, if I pushed them right to within a few percent of their 'limits' ended up having those limits steadily decrease as the cards aged. My old i7-4770k that's been at 4.4ghz for over a decade finally succumbed to it too last year after refusing to POST at settings that worked for years. After redoing all the stability tests I did back when I first tuned it, it's now only stable at 4.2ghz with a 75mV Vcore bump over its relatively modest previous settings, after ~10 years of use.

It's silent and transparent to the user, but modern CPUs even do this out of the box. AMD's boost algorithms have been adapting and adjusting for aging/degradation for years, either by reducing boost clocks, increasing the voltages required for those same boost clocks, or both: https://www.anandtech.com/show/1487...ing-perception-with-amds-frequency-metrics-/3 (close to the bottom)

 

[arandomguy]

Another aspect of this is that "stable" OCs even right at the onset are really only tested stable under the specific conditions that they were tested at. You change any of the variables involved (and there's tons) and it can change the probability of an error occurring.

In general there's reasons why there is so much clock and voltage headroom (even nowadays). It's not because of some conspiracy, agenda or ineptitude as some online comments would suggest.

 

[Albuquerque]

There's no question silicon can and does degrade, however it was a very distinct change during the major driver change. I was able to demonstrate it later by rolling the driver (cleaner and all) back to 560-something and the prior OC was back to being completely stable. Also consider 950mV (and even 975mV) is quite the undervolt for a 2700MHz clock for the 4090 -- or at least mine. For the most part, my card (when fully loaded) wouldn't even hit 2700MHz for more than just a second or two at stock voltage -- which was 1050mV at that. I'm not reasonably expecting silicon degradation with a 10% drop in voltage and quite a bit less in total power consumption.

Second, it's very much true stable overclocks are built upon a consistent ecosystem, and changing drivers is absolutely a reason for an overclock to become unstable. Typically, in my nearly four decades of experience with PCs, I'd expect a driver update to increase performance while also making my overclock unstable. A better optimized app or driver can absolutely increase the load on the hardware. The sad thing is, I'm not sure the 570's materially affected the performance of any of the games I would play with them.

Regardless, it was a difference, and yet somehow didn't result in blackscreens lol...

 

[Sega_Model_4]

Has anyone here had these issues?

Even though I'm on the RTX 30 series, these drivers are also causing me problems. I was thinking the problem was the game, Dragon Age Inquisition, being from EA, a massive RPG, it would be prone to bugs.

If I'm playing the game and turn off the TV (OLED, right? I don't risk leaving the TV on if the game is going to be paused for a long time), the game crashes. Since Gamer Nexus mentioned issues involving monitors, frequency, Gsync, I assume it could be related. And in the most recent driver, the game started to have some graphical bugs, with the geometry sometimes taking a long time to render (like PS1). Since I've been playing this game since January, I hadn't seen these bugs before, only now in the most recent playthroughs (like yesterday).

I'm going to revert to the December driver to see if anything changes.

 

[DegustatoR]

Hmm...

Many months ago, thinking at least six and probably closer to nine or 12 months ago, my Monoprice 35" VRR 100Hz 3440x1440 DP-connected monitor starting having blackscreen issues every once in a while. What I eventually discovered after fighting it for a while was VRR support was somehow b0rked. I disabled VRR support in the driver and I've had literally no issues at all with my monitor ever since, and I'm 100% confident this started in the 560-series drivers. What's interesting is, at least for me, I've been keeping up the 570-seriies driver updates and have yet to (re-) encounter any blackscreen issues.

I have absolutely hit a few strange instabilities with my long-standing stable overclocking profiles in the 570-series drivers. My 2700MHz @ 950mV setting has been stable ever since I bought my 4090, yet somehow was no longer capable of starting a Cyberpunk 2077 saved game without crashing to desktop. I eventually re-settled on 2700MHz @ 975mV which is fine, yet is somehow measurably different from where I started.

If that happened around 24H2 release then you've probably got MPOs enabled in 10/12 bit color and these do expose VRR issues on some monitors as MPOs can trigger refresh rate change for the desktop when you open some apps which are using GPU acceleration. Can generally be fought with by disabling said acceleration in these apps - or disabling VRR.

 

[Scott_Arm]

If that happened around 24H2 release then you've probably got MPOs enabled in 10/12 bit color and these do expose VRR issues on some monitors as MPOs can trigger refresh rate change for the desktop when you open some apps which are using GPU acceleration. Can generally be fought with by disabling said acceleration in these apps - or disabling VRR.

Cool shit ... Is this a windows issue or an Nvidia implementation issue?

 

[DegustatoR]

Cool shit ... Is this a windows issue or an Nvidia implementation issue?

My guess would be monitor VRR h/w issue?

 

[Albuquerque]

Honestly, I haven't much missed VRR ever since I discovered it was the root cause of my monitor blanking issues. Nearly all the games I play can hit the 100Hz refresh rate of the monitor, albeit with some help from DLSS and on rare occasions some FG, all of which makes VRR mostly irrelevant. It wouldn't surprise me to learn there's something slightly off about the VRR on my particular monitor, given what little I paid for it. Despite the low cost, I've been very happy with it and have no specific intentions of replacing it anytime soon.

 

[SlmDnk]