NVIDIA GeForce 497.29 WHQL driver download & Discussion

What do you aspect from the January drivers over this new driver?
If they not work well you go back to 472.12 again ?

 

I always have Hags on I never had trouble with it ,I am also on windows 11

 

No you have to disable it when the installer starts you can choose it to not install it ,if you don't you can't uncheck it anymore

 

i have no friggin idea how people are reporting lower mhz with every driver.that from my experience is just impossible to replicate.
i have a stable mhz/mv curve set up in msi afterburner,and clocks remained the same across three different driver branches I ever had installed,with no impact on stability.tested on metro ee,a game that is so prone to system instability that it'd crash if you look at the computer wrong.

it's a worthless turd anyway

 

So far so GooD, great, stable & performant driver.

 

497.29 like every 497.xx brings on my setup 3-5% lower fps comparing to 472.12

Win11, Z690, 3090 Strix

 

I wish Nvidia or SNK released a fix for King Of Fighters 14. This game keeps crashing after 2-3 rounds of fighting. I've done all I could possibly do to get this game fixed but myself and so many others have hit a brick wall. I'm starting to think it will never be fixed. 3000 series cards can't play this game without crashing.

 

Hello @ManuelG ,

Same problem as mine : https://www.nvidia.com/en-us/geforc...rds/5/462767/audio-drops-on-46647-and-higher/

Thanks.

 

weird, i´m playing this game normally, just a little artifacts in special intros

 

this is what I found.

same issues as prior dch 496/497
Same visual polygonal small black manifestations around open windows in 2d desktop. Same issues with dp 1.4dsc and hdmi 2.1 on 4k 10 bit. Same performance (within margin of error) as dch 496/497. Issues resolved by reverting to standard 472 driver or below (461-472). No decrease in performance compared to 496/497, but overall decrease since 466-471.


issues reproducible on multiple machines running similar configurations.
I have tickets in with nvidia on visual artifacts and the dp 1.4dsc and hdmi issues from the 496 driver incept

I didn't see any reason to copy the benchmark results again in this post. I feel like I'm beating a dead horse.

The basic observation, (beyond the visual and operational anomalies), performance is degraded on high demand games/apps when full eye candy for max accuracy visual is enabled, no DLSS and no film noise, no DOF ,no vignette and no motion blur at 4k+. This trend is consistent from the 460.xx series drivers to the 497, with the largest drop of in performance being the 496.497. So maximum fidelity at 4k+ has a consistent performance drop.

High resolution, max eye candy and NO DLSS has shown a performance drop in dch 496/497 compared to standard.471.12, 471.22, 472.47, 466.77, 461 etc.

 

It's the last NON DHC driver right?(Currently on that driver but using standard)

Coincedence?

 

The next one will be a patch for the hotfix that patched the patch. This is what happens when drivers follow the Micro$lop model. A constantly evolving state of brokenness.

This is a common, but still clever, way of slipping in PUPs and malware bundled with legit software.

 

Yes.

 

472.84 is standard, @venturi reports similar issues with those.

472.47 standard matches / is close to performance of the 472.12 driver.

See post comparing 461.xx, 466.xx, 471.xx, 472.xx, 496.xx at https://forums.guru3d.com/threads/nvidia-studio-driver-v472-84-whql-released.440967/#post-5973908 .

 

What do you expect if they are adding security fixes to almost every new drivers release

Since I have RTX 3090 I didn't notice better performance thanks to the drivers, actually slightly lower results which might be due to those security fixes.
On my second PC with RX 6900 XT new drivers can improve performance, even if numbers are only 4-8% better its nice to have, especially in 4K when every single FPS counts

 

I wonder how realistically easy are these "security flaws" to fall victim too.I really hope they're not "fixing" these issues at the cost of performance when the only way you're gonna fall victim to them is by some really wierd way like actual meddling physically with the gpu.

Does anyone actually know someone who got compromised trough their gpu?

 

Vulkan
461.72 driver - Gravity Mark score 71.382
466.77 driver - Gravity Mark score 71,424
471.22 driver - Gravity Mark score 71,430
471.41 driver - Gravity Mark score 70,663
472.12 driver - Gravity Mark score 69,057
472.39 driver - Gravity Mark score 68,841
472.47 driver - Gravity Mark score 69,985
496.61 driver - Gravity Mark score 64,362
472.84 driver - Gravity Mark score 65,554
497.29 driver - Gravity Mark score 64,398

DX12
461.72 driver - Gravity Mark score 67,344
466.77 driver - Gravity Mark score 67,211
471.22 driver - Gravity Mark score 67,192
471.41 driver - Gravity Mark score 67,315
472.12 driver - Gravity Mark score 66,758
472.39 driver - Gravity Mark score 65,446
472.47 driver - Gravity Mark score 65,888
496.61 driver - Gravity Mark score 60,271
472.84 driver - Gravity Mark score 62,003
497.29 driver - Gravity Mark score 60,417

https://gravitymark.tellusim.com/leaderboard/

...as close to pure test environment, I boot into this build to make it apples to apples.

testing parameters and observations on apps/games:
4k, monitor at 10 bit, NO DLSS, no DoF, no motion blur, no vignetting, no film noise, room temps about 69.5F - 71.5F,

Hardware build and OS image used are unchanged throughout the driver score comparison.
Driver is installed as "cleaned" no GFE, etc, only driver and physics.

System specs:

2x 3090 RTX Founders Edition & SLI / NvLink bridge
2x 8280L, 56/112 cores, Asus c621 Sage Dual socket motherboard
1.5 TB ram. DDR4 ECC LRDIMMs
1600W silent digital power supply
(Data drive) 4x VROC Raid 0 Micron 9300 Max (12.8TB each / 51.2 TB volume)
(OS Drive) Sabrent Rocket 4 Plus (4TB). 4x Phanteks T30,120mm for CPUs, 92mm for nvme drivers
Asus PA32UCG-K monitor,
TT SFF case, micro. and heavily modded
MS Data Center 2022 & Ubuntu

https://forums.guru3d.com/threads/dual-3090-rtx-dual-xeon-sff-case.440516/



Removing most variables, apples to apples, high demand apps /games, 4k max eye candy, no DLSS, no system bottlenecks, no hardware bottlenecks, as pure as I can make the OS, no overclocking, with consistent room temps and consistent settings (results are consistent with the other similarly configured hardware systems in my house)



----the results shown a decrease in performance (in my environment) Anomalies and performance issues encountered on dch 496/497 seem resolved only by reversion to standard 472.47 (and below) driver series.

 

I was given a 1440p 165Hz gaming monitor for xmas, so it's the first time I am running 2 monitors. My old monitor is a 1080p 75hz monitor. What I have noticed now is my GPU (1080Ti) runs at full overclock continuously. Is this normal? I have done a lot of searching and there seems to so many different answers. Even Nvidia claim this is no longer an issue with newer hardware and drivers. I tried the tool in Nvidia Inspector, but then games had issues as the graphics cards where not switching back to the correct power states. I have only used the 497.09 and 497.29 drivers since getting the additional monitor.

 

2K result?
Regardless of tweaks and overclocking, 4K and 8K results should have been beaten by older drivers?

GravityMark Report (tellusim.com) Top result 4K Vulkan with driver version 469.13
GravityMark Report (tellusim.com) 4K DX12 471.41
GravityMark Report (tellusim.com) 8K DX12 496.13
GravityMark Report (tellusim.com) 8k Vulkan 496.13

With more Asteroids fps seems to be more aligned with what is to be expected in games, 496.13 seems to be on top for that particular benchmark there too.
That aside, I rather look at real gameplay results with clockspeeds, fps, frametime graph and none the least, GPU temperature, so taking all these benchmark results with a grain of salt.

Statistic of how much difference you get in actually gaming sessions without pushing the GPU/GPU's beyond 100% load is way more interesting to know and might also be useful for Nvidia's driver developers.

Nice build and monitor you got there btw.

 

Flicker is worse with these drivers last four sets of drivers have being doing this.