GeForce 466.47 WHQL driver download & discussion

There was a possible fix, postet by Astyanax (thanks!) a few drivers ago. Try to use PhysX V 9.20.0221 - I have it installed and no problems with Arkham Night (turbulence smoke enabled)
Here ist the original post: https://forums.guru3d.com/threads/n...driver-download-and-discussion.435609/page-15

 

I confirm that it is a driver problem, putting the YCBR442 audio format at 10bpc the slowness is excellent, but putting 8 bpc rgb etc. high slowness. can anyone open a post in the nvidia forum?
lg oled 48C rtx 2080ti HDMI

 

it does't affect Turing and Ampere.

 

Quoting so i don't have to link every post

@tty8k can you see if this continues to occur with youtube under chrome canary?

90 and 91 are both still affected by the bug where the surface is changed every second.

 

Please make sure to fill out the NVIDIA driver feedback form:
https://forms.gle/kJ9Bqcaicvjb82SdA

I usually check the driver feedback form first before I check forum feedback as it has more complete information that I can use for filing bugs.

 

Hopefully we get this fixed now.

 

I had some stuttering in games in the past couple of drivers. I changed the output from default 8 bpc/Full to 10 bpc/Full in Nvidia Control panel, and the stutters are gone. There is definitely something fishy going on with 8 bpc in these 466 drivers that for whatever reason hampers performance. I didn't check DPC latency, but I'm guessing it's related to what others are reporting.

 

wouldn't it be worth opening a post on the nvidia forum? They have the idea that they do not longer a good driver

 

Manuel himself responded here so I believe its being looked at now.

 

Funny how people are reporting Latency issues in the drivers, which I've been reporting for months... All I got was "It's your system". Now it looks to be an issue with the colour settings of 8bit

 

It is in part your system (an aspect of atleast), thats why you were told to send them trace etls and diagnostics, not everyone with 8bit is experiencing this, its certainly not occuring on pascal, not even all turing and ampere user sees it either.

Theres a condition (apart from the 8bit setting) shared by some of you that others do not - and its looking like 9th gen intel processors may be it (if not a bios / microcode concern)

 

Definitely not, promotion spikes across 500us do happen but going above 1000 is really excessive.

with all Cstates on the driver might see a +600 spike and be 250 and lower averaged (or below 100 without c states), the 1500us spikes are way outside the norm

my 1680v2+1060 are 228us all up with all supported c states/core parking on(balanced), 100us with high performance plan(but c states still on)

I can get even lower with c states off as well.

 

Thank you

 

It's not exclusive to 9th gen intel (I have ryzen 5800x) and not only 8bit, but HDR is affected as well. Only when selecting "use nvidia settings" and forcing 10bit desktop the 1200us spikes stop.

 

Even tho turning to 10bit seems to fix the issue.An actual fix from nvidia would be nice.Hopefully we get that.

 

Oh well, I'm stuck on 8bit and can't change to 10bit to test

 

Yep you're kinda stuck :/
I really hope they make it a priority fix but with nvidia you can never tell.They have yet to fix VR stutter after all this time so I don't really have any faith in them.

 

my cousin in the same situation. what a shame nvidia
i7 9700k, 2080 ti Dp.

 

460.89 works fine on my system.

 

I've asked multiple friends to check this on Ryzen and they don't experience it with 2070 super / 3070.

This..... does bring it in circle back to a behavior that cpu's with iffy pcie controllers don't like. -potentially.

i'll keep asking around.