Nvidia Hotfix Driver 496.98

These drivers doesn't affect your sense of humor LOL.
Driver works fine here too

 

No, DX11 and DX12 running fine here and i didn't notice any performance issues so far, this driver is a good driver IMO.

 

You could try driver 466.77, which according to his tests seems to perform better overall. I mean, his list is pretty clear, is not that you should sit on "some 460" driver, use the one that showed the best results there.

 

Except it isn't that good. Doesn't mean anything if you think it is.

 

I really hope you guys don't extrapolate having to change to an older driver solely for the reason that someone with an extremely exotic hardware setup (and even just SLI for that matter) posted worse benchmark numbers in a benchmark software almost no one uses. He himself mentioned that, too.

Users like RodroG or lokkenjp spend hours upon hours benchmarking various games and synthetic benchmarks and have never found any long-term degradation in performance.

 

To be fair, my issues are not reported as SLI, I mentioned SLI as part of an enthusiast forum. I am not having ANY SLI issues.. Some apps are DX12 or Vulkan MGPU, which is supposed to be supported, the degradation occurs in MGPU mode on 3090 RTX FE as driver builds progress- when the only change variable is the driver

Things to consider: performance degradation occurs in the 4k plus range, 10 bit, max eye candy, etc. and may not demonstrate itself at middle resolutions and system demands
Example, performance degradation in gravitymark, is prevalent when the number of rendered objects increases (2 million per scene) at 4k. In that case there is a considerable degradation from the 471 driver to the 496. Under normal games and average/mode configurations, there may not be any measurable change, which most folks would never notice. The details of the issue are what makes broad stroke analysis of the situation blur the problem.


The actual issues I am indeed reporting are the black polygons on DCH install, the DP 1.4 DSC and HDMI 2.1 response time on the 496.xx drivers and the alt tab 1-3 sec black out when alt tab from 3d app/game to desktop when using 10 bit mode and the monitor refresh rate and setting is the same as the desktop.
Those are the issue/problem of reporting. yes, I actually have a ticket(s) in.

Lastly, yes , for a high end configuration, I am experiencing performance degradation when the only variable of change is the driver, ---but my main request is for the other items I reported to be solved, which seems a reasonable expectation

 

maybe not for you ,but for me the driver is ok, if you can see what i think would be a miracle pfff LOL

 

I tried 510.06 driver and i found it good and stable

 

It isn't remotely stable.

 

So far so good, played fh5 and victor vran for hours all good, i lost like 2-3 fps in game but i feel its more stable ( less stutter also), keeper until a new driver release

 

This driver is messed up. 2nd time in 3 days my DP monitor suddenly stopped working and DVI monitor (usualy turned off) tooked his place, DP monitor is nowhere to be seen in settings, only restart helps.

 

To state clearly what it is: Gravity Mark nvidia driver performance degradation @ 4k? 8k? and 10bit

As for the rest, got some actual game benchmarks over all those sets showing the exact same trend?
I also take it as that this is when pushing GPU load to 100% or unplayable?

 

Hmmm no shader cache size selection in the CP on this driver, but it is indeed stable.....

 

Yeah after installing a driver update it is the best to reboot and not only for nVidia drivers.

 

Don't know if this is specific to this driver but I recently updated Star Citizen to the latest live build and experienced a lot of hitching where the game would just stop for half a second as if to load content on a slow drive ( game is installed on Crucial P5 Plus SSD ). I saw a post on reddit that said to up the Shader Cache to 10gb to fix it and it did make things better. I also noticed I was experiencing similar stutters with the latest live version of Satisfactory ( in DX11 and experimental DX12 ) and since I upped the cache size those stutters are now gone. This was never an issue in the past so not sure what might be happening. Could be a game issue, could be the fact I'm on Win 11 or something else ? In any case, with Shader Cache set to 10gb all seems well for now.

 

There is some funkyness with DP with this driver here as well but I'm not entirely sure just yet that it's the driver fault.

 

Not seeing these issues on 497.09 so far so it may have been a driver issue after all.

 

I'm experiencing the issue, but its very specific to a certain criteria, not random

not just DP but in DP 1.4 DSC when set to 10-bit specifically and at 120 or 144hz

I've noticed a concurrent side effect, After 472, when the issue manifest, I've also notice that in control panel it reports my 10-bit monitor as a 12-bit, moves the resolution from the " PC 3840x2160" section in control panel to the " UHD 3840x2160" section.

AND
it moves the monitor refresh rate from 144hz or 120hz to a "119 hz".

Make all the changes back the way they should be and upon reboot all the wrong parameters are back.

Issue started with 496 series.

497 is doing it as well, but at the same specific level.




I don't know if its driver or DCH, so its not a DCH haters comment, however, up till 472, I only used standard and I'm experiencing this issue as well as black polygons on 2d desktop ever since the 496 and 497

These issues go away when reverting to 472, 471, 466 or 461 standard drivers.



Hope this helps

 

Not the same issue.
I was losing connection over DP 1.4 10-bit 4K 144Hz.
It was a second long blank screen and a Windows logon screen after that.
Happened a couple of times each time during the first hour after the display was awaken from a long standby.
Doesn't occur on 497.09 so far.

But in a related note the driver reports that a new Gsync display was connected after a long monitor standby periods.
It's not a big issue since that's all what happens really (res, Hz, ICC are all correct) but I wonder what's so "new" about a monitor which went into standby and then went back to "on" again in a couple of hours.

 

sounds like the behavior of the displayport standards deep sleep capability.