Discussion in 'Videocards - NVIDIA GeForce Drivers Section' started by norton, Apr 11, 2019.
The burden of proof is on you my friend.
And the 2080ti.
in different rigs around the house.
I happened to have some old Heaven BM data from when I first got myGTX970 (2014), I ran the BM again with the last driver version and was almost the same, within a few points.
So no gimping that I have seen.
Unigine Valley 4k, Ultra settings NoAA - directx11 backend
399.24: 2656 points - min: 29.3 / max 115
425.31: 2639 points - min 27.4 / max 110 <== difference for min frames = 7% / difference for max frames: 4,5%
Where the fps starts to count the new drivers do slow down my gtx 1080 up to ~8%
The same story is for the Heaven benchmark.
As for the Rise of the Tomb Raider I could see a 2,3 fps drop by comparing the same scene.
AND these are not small numbers as we are always trying to get the best from our cards.
Where the scenes are becoming very complex the slow down is starting to be noticeable as I get a 8% performance drop - such 8% can be easily rounded to 10%
There are people out there that are throwing their money to get that 10% performance boost.
Czesc Densou W rescie troche wsparcia w tej walce
Guys, I understand that many of you do prefer to get the info from *legit* sources.
For me it is ok, I wanted only to say hi and let know NVIDIA (cause they are reading this forum as well) that someone will not buy their products any more - easy right?
1080 GamingX OCed. Timespy
399:24: 8146 GS
425.31: 8362 GS
Great Driver on BFV dx12.
I ve started to play with hardware when Linus was still learning to say "mum"
Please, do not try to smash in front of my eyes those people and those sites - it does not impression me much
If people like me that have always liked the green company (even if we all know that they are not saints) are complaining about the 4xx drivers performance for the past generation of graphics cards IT MUST ABSOLUTELY mean that something is not right.
No dx12 here, sorry.
It seems your mind is gimping too! no one care about of benchmark numbers because every driver has it's own improvements in specific case.
So happy that your mind is 100% functional...please read carefully my previous posts...try to repeat the process to be really sure.
here my data on few drivers on GTX970
Heaven D3D11, 1600x1200 8xAA FS, Prest-Custom, Quality- Ultra, Tessel- normal
FPS 74.2, score=1868, Min fps=29.2, Max fps=140.9
FPS 72.5, Score=1827, Min fps=29.9, Max fps=137.7
FPS 72.8, score=1835, Min fps=30.5, Max fps 137.6
Pretty consistent results across a broad timeline, though the latest happens to be best it seems.
In games I play I never really notice any difference.
Edit: your results are under 1%, about 0.75%
399.24: 2656 points
425.31: 2639 points
Michael, you are providing numbers that are in the range of being affected by atomic level variables.
You want so desperately for there to be a conspiracy that you are creating one.
no sound now at all after installing these drivers lol;
What does your claim about your experience have to do with anything? I could claim I started on hardware older than your dad...and so what? Your singular claim of 'gimping' is countered by literally millions and dozens of online tech site benchmark who have seen nothing beyond the margin of error (disregarding the occasional regression like Witcher 3 for example which was fixed.) the performance is the same or better. Frankly the adults get bored with what can only be trolling for attention given how well the facts are established.
I'm sorry but "people" "claiming" anything on forums means exactly zero to me since for all I know these "people" are those who has been paid to "claim" whatever they are claiming. There are benchmarks which anyone can run and compare. No need to "claim" anything, just post results. In 99.9% of cases where there are "issues" it's a user error and little more.
Check your sound mixer. Sometimes it changes the default device for no apparent reason when installing new drivers.
From my side there is no driver passing version 399.24 that allows my gtx 1080 to get the same fps as with 399.24.
And when we speak about moving from ~30.0 to ~27 for sure we can not imagine to stand out of the masses with some tangible revelation.
But, those marginal differences do make many of us happy, especially when we are usual to play in 4k and we do not bother to have synced the vertical refresh rate up to 60 frames or higher.
The *new* drivers start to slow *marginally* my gtx 1080 at 4k when the scenes start to become complex.
And passing from 30 fps to 27 is not so sweet.
I can understand that there may be some factors that can influence the results just like some overclock yielding hidden somewhere in the new drivers but the new ones should preserve all the goods that were part of the previous versions, oc included.
I for sure can make a recheck without any oc but this will imply to install once again the new drivers and then get rid of them as I can not live without it.
And before making all that noise here - please see that my real name is visible to everyone (I did not hide it during subscription and there is no sense to do it now) - I did ponder if it was the case to.
It is bad thing to let a fan/supporter of a company to make it disappointed with some questionable practices.
It has: before firing I do wait a bit more than a young soul.
Attention and trolling you say.
Trolling is excluded as I have a nvidia gtx sticker on my case and seriously I always have had a good reputation of the green guys, till now.
Attention - sure but not yours.
You said exactly nothing but to imply somehow you're mature and have experience. That you have a sticker has nothing to do with the content of your posts. The facts as I posted them are what matters not your opinions son. I have the facts you have your musings on your age and experience. Your post would imply you have not taken advantage of either. You're factually wrong and we know you're factually wrong. The end.
I usually post in the nvidia cards thread. I have detailed threads on SLI, titan Xp quad SLI, Volta, and Titan V. These are my observations.
I don't have information on the 399.24 driver, however I do have some data points on the 399.40 driver. I am on volta, and as such any drivers in the 400 series to date have been 1-7% slower in fps than the 399.40 on dx12 games. I have 4 volta cards in my computer and they are used for NN and medical imaging. I game, but I am not an "avid" gamer.
Beyond the up to 7% lower fps, I have other anomilous behavior such as instabilities and hitches on any driver version in the 400 series. I have several different loads I can use. I have tested this hypothesis on win 7, win sever 2016 data center, windows 10 (blah), and windows 2019 server.
The drop and instability seems to manifest across multiple platforms. However, there seems to be less issues on the win 7 load for instability, also obviously on performance there is no dx12 games on win 7, and issues seem to be relevant to dx12 games.
up to driver 416, I had in my possession 4x RTX titan, but fund them slower and less capable than the volta for my uses, so I resorted back to my volta environment. So I cant't objectively include rtx observations in this post as it is for the 425.31 thread.
The problem seems more apparent on dx12 games. So in conclusion, here is the detail: I see a pattern on volta (I can only speak on volta), maybe not as strongly as yours, that dx12 games have taken the 1-7% fps hit (and instability) on driver series 4xx as compared to the last 399 driver, the 399.40 specifically. The 425.31 driver manifest the same observations.
exceptions to observations:
Vulkan seems to have gotten better, but I also manually add vulkan updates. The 399.40 with updates performs on par.
Ashes of the singularity dx12 mgpu has gotten better on 400 series compared to 399.40 so this is a true exception. It also happens to be the ONLY game with real mgpu (to date).
Strange brigade has gotten better on vulkan mgpu 400 series driver (3%)
Strange brigade dx12 mgpu dropped on 400 series driver (-7%)
UE4 Echo ultra ultra has stayed the same in perf/fps, but load times for levels has increased on 400 driver, also on occasion on 400 driver series the game crashes and exits to desktop, this behavior does NOT manifest on 399.40 (on volta and server 2019 data center)
Agony, also UE4, has manifested the same behavior on 425.31.
399.40 is a stronger and stable performer (on volta and server 2019 data center)
416.30 modded inf quadro is the better of the 400 series driver so far (on volta and server 2019 data center)
425.31 displays many of the negative characteristics of 411, 417, and 419 on volta
the driver in this thread (425.31) seems to have a lower performance (1-7%) than 399.40
(on volta and server 2019 data center)
the hardware used during observations:
Dual 8180M xeon, asus sage 621e board
384GB of ecc ram, 10x ssd raid, and 960pro nvme
4x Titan V, including 32Gb CEO Ed.
(I also can turn off/on sli using auto diff sli, testing was done without the modified driver)