[Boards: 3 / a / aco / adv / an / asp / b / biz / c / cgl / ck / cm / co / d / diy / e / fa / fit / g / gd / gif / h / hc / his / hm / hr / i / ic / int / jp / k / lgbt / lit / m / mlp / mu / n / news / o / out / p / po / pol / qa / r / r9k / s / s4s / sci / soc / sp / t / tg / toy / trash / trv / tv / u / v / vg / vp / vr / w / wg / wsg / wsr / x / y ] [Home]
4chanarchives logo
Did some driver testing to confirm whether or not Nvidia has
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.

You are currently reading a thread in /v/ - Video Games

Thread replies: 55
Thread images: 7
File: eTT4zz7.jpg (804 KB, 1920x1080) Image search: [Google]
eTT4zz7.jpg
804 KB, 1920x1080
Did some driver testing to confirm whether or not Nvidia has been deliberately gimping Kepler cards in recent drivers, here's the results.

Tested in Arma 3 since the game is known to not be optimized and not favour either of the brands as it's hard to find a game nowadays that isn't Nvidia or AMD afiliated.

All tests were performed by loading the game at the exact same spot and looking at the horizon for roughly a minute to stabilize the framerate so it would be a constant.

The only thing that changed between each test was the driver revision, all on default settings with clean install option so no tweaking of any sort being applied to drivers, just clean drivers exactly the way they come.

There was also the Witcher 3 performance fiasco shortly before 353 driver launched but Arma 3 performance doesn't seem to have been affected, the "Kepler Fix" Nvidia claimed to do was in driver 353 which bumped performance up in Witcher 3 but none of that seems to have had any effect on Arma 3 and it's safe to assume it was Witcher 3 exclusive fixes.

Anyhow.
The first entry are the first major driver release since 780 Ti came out, basically the "entry" drivers for the card with which it more or less launched with.
Then it showed gradual improvement all the way up to 353 series.
Starting with 355 there's an observable subtle downward slope in performance that seems to be going down roughly 1 frame with each new major driver. Over the course of less than a year since 353 drivers, Arma 3 exhibits a performance worse of 5 frames.

Note that all this was tested on the exact same version of Arma 3 which is the latest build with new improved renderer.

I intend to do more tests in some other games like Witcher 3 and Dark Souls 3 next.
>>
File: 1446929530464.png (106 KB, 1008x1216) Image search: [Google]
1446929530464.png
106 KB, 1008x1216
>meanwhile, at AMD
>>
>>335832684
You're not me so stop shitting up the thread with fucking AMD shilling.
I never owned AMD card in my life, I've got 780 Ti as you can see and I only did the tests to verify if there actually has been a downgrade.
>>
>>335832308
Doing gods work anon.
>tfw I never updated from 347
>tfw uninstalled geforce experience since it fucked up all my settings in inspector

Dodged a bullet there I guess.
>>
You should post this some place else as well for it to gain some friction.

Hell, you could even start a fire at Plebbit with this probably.
>>
Is there a way to determine which driver is the "347.88-353.62" of other video cards without doing this whole suite of tests?
>>
>>335833007
Can you be more specific? Because I'm afraid I don't understand your question.

>>335832938
I don't even know how to use that site so feel free to repost, I don't even need credit or anything, I mostly tested this only for myself.
Mind you this is only 1 game so it doesn't mean it would apply to every one of them but it does raise some alarm in my head.
>>
>>335832308
do you have a DDL link for the good driver?
>>
I've been using Nvidia since 2008 and I have never installed the Nvidia Experience crap, it just reeked of problems since day 1.
>>
>>335833390
My point is that this thread will disappear at some point in time, so you might want to post it somewhere at there very least where it stays up.
Would be a shame for this effort to go to waste.
>>
Take OP's post but with a grain of salt.

What qualifications does he have that would lead him to come to such obtuse conclusions?
>>
>>335833438
Those tests were done WITHOUT Nvidia experience. Just clean drivers.

>>335833583
Like I said, you're free to repost it. I actually intend to repost this existing information after I do more tests in other games.

>>335833404
https://www.nvidia.com/Download/Find.aspx?lang=en-us
Just find it here
>>
I recently had to rollback my Nvidia drivers to get my GTA V fps back to a reasonable level. Rolling back drivers more than doubled my framerate, while allowing me to boost the quality by a lot, too. Either they fucked up something royally trying to get one particular game working / working better or they are desperate for hardware sales even though only like... 5 dev teams can push hardware legitimately anymore, if that.
>>
>>335832803

Roll with it amigo - according to /v/ and /g/ AMD have no drivers so for hawaii to now be annihilating big kepler in modern games it must be a driver gimping right?
>>
>>335833646
Are you impying you need special training and qualifications to literally install a fucking driver and boot a game to test framerate with any software that displays framerate? Are you serious anon?
>>
>>335832308
I'm on a 770 and am on 350.12

>>335833758
>>335833646
No, you should. Even on the exact same hardware stuff can be different between two PCs. Still it's not always drastic
>>
>>335833758
I'm 'inferring' that your single point of data isn't enough to make a convincing assessment.
>>
>>335833956
It's a relatively clean Windows 7 Ultimate x64 install from 2 months ago.
Other important hardware specs:
8GB Kingston HyperX DDR3 RAM
i7-2600k overclocked to 4.0 GHz

>>335834083
By single point you mean one spot or one game? I already said I intend to test more games and that nobody should go apeshit just yet as it could be a case of optimization for some other game fucking over driver calls for this particular one.
>>
>>335833707
What driver were you on and went to? Apparently the GTA V driver 350.12 is really stable for most people. That's why I use it
>>
Good tests OP. Nvidia are scam artists.
>>
>>335833390
I'm basically asking if there's a way to find out what the GOAT driver is for other video cards that doesn't require doing extensive testing and documentation such as this one. For example, I own a 560 Ti and I know its GOAT driver is 314.22. But in the future I may want to pick up a new videocard, but I have no idea at which point are the drivers starting to go downhill.
>>
>>335832308
You need to make a video of this. Your words alone isn't enough proof.
Post it on youtube and hopefully people will spread it.
>>
>>335832308
>reddit filename

AMDshills strike again
>>
>>335834231
I went back to 362.00... not sure from what

running fine
>>
>>335834373
There is no other way other than doing extensive testing unfortunately.

>>335834394
>Your words alone isn't enough proof.
That same logic can be applied to anything I show so if somebody doesn't believe I honestly don't care, did this mostly for myself.
Also here's an album with the tests in case someone doesn't believe the collage pic:
http://imgur.com/a/tojgm

>>335834636
It's Imgur you retard. Do show me a better hosting site for sending pictures to friends and I'll gladly swap.
>>
So where's the proof that these are the actual frames that were displayed? You should've made a video or something, this just reeks of shitposting
>>
>>335835168
>this just reeks of shitposting
Thanks for not reading, again.
I genuinely regret even bother since you reads keep jumping to conclusions instead of actually fucking reading
>>
>>335835285
>BELIEVE ME BECAUSE I SAID SO REEEEEE

nah bro, you can either supply substantial evidence or get the fuck out. It's pretty telling that you got super defensive when I pointed out your obvious shitposting
>>
I usually update, but i've stopped a couple weeks ago, since anything beyond 364.47 just destroys any functionality from my GPU. Everything just goes on black.

770 btw.
>>
>>335835536
You're the one literally shitposting because you refuse to fucking read, see >>335834868 for evidence you dumb little shit, and of course I get defensive when a literal cockmongler refuses to READ THE FUCKING OP
>>
File: get.jpg (62 KB, 728x546) Image search: [Google]
get.jpg
62 KB, 728x546
>>335834652
That's what I'm on as well, the 364.xx drivers are absolutely fucked beyond recognition. I was getting frame pacing issues in any game with HBAO turned on.

Btw for all of the people asking for "proof" or a video, check this video out. This guy's got to be part of the industry.

https://www.youtube.com/watch?v=O7fA_JC_R5s

Most of the problems are surrounding gameworks.
>>
File: 011.jpg (216 KB, 1600x960) Image search: [Google]
011.jpg
216 KB, 1600x960
>>335835536
Not OP but how's this for a proof nigger?
https://www.youtube.com/watch?v=hImoox6qyzo
https://www.youtube.com/watch?v=eKk63TajS6M
>>
>>335835725
Does shadowplay not work on that card? I don't understand why'd you use bandicam when shadowplay is way less intensive
>>
>>335834868
>That same logic can be applied to anything I show so if somebody doesn't believe I honestly don't care, did this mostly for myself.
A video taken by your phone would hold up ALOT more than just your words and some pics. I'm responding because I would love to this snowball into a shitstorm, but it's up to you.
>>
>>335835725
Wow that's nuts.

Who still has a GTS 450 though?
>>
>>335835624
How is an album of images any better than a single image? You're clearly just mad as fuck you got caught out shitposting, the real question is whether you're doing this for amusement or if you're being paid.

>>335835725
thanks, i'll watch these
>>
>>335835725
did you notice that as soon as he hits an area with a lot of tessellation the framerate nosedives?
>>
>>335835939
Get cancer and die seriously, see fucking >>335835725 for more proof you literal Nvidia cockslurper.

>>335835901
Yeah I'm gonna record some videos now because of this cockmongler >>335835939
>>
File: expect me to read.jpg (51 KB, 392x500) Image search: [Google]
expect me to read.jpg
51 KB, 392x500
>>
>>335836060
Cool story bro, I'm sure that you'll deliver. We'll wait
>>
File: 76b0742d78.png (34 KB, 379x453) Image search: [Google]
76b0742d78.png
34 KB, 379x453
sure glad i dont need to downgrade drivers for specific games every time i wanna play something different
thanks amd
>>
>>335832308
Do it again using DDU to wipe the drivers. The clean install option is NOT a clean install, it just resets nvcp options.
>>
>>335836529
This was with the DDU, namely the one from here
https://www.guru3d.com/files-details/display-driver-uninstaller-download.html
>>
Can you play newer games with older drivers? Like if I want Mankind Divided in August, can I play it on drivers from 2015?
>>
>>335836653
Yeah. Just wait to see how MD's drivers are for your series of card first.
>>
>>335836653
You can.
>>
>>335836653
It depends on whether or not the drivers add features that aren't supported in older versions.

Otherwise, yes.
>>
>>335836613
Nope. You never mentioned DDU in OP, everything was detailed and you clearly thought the clean install option was an actual clean install.

Have fun with your troll thread.
>>
>>335837109
Nice projection asshole but alas you're wrong. Go shill more for Nvidia please.
>>
>>335837109
And you can prove that op didnt use ddu how? Sorry but i dont see how your claim is anything more than an empty claim
>>
>>335836060
Glad to hear friend
>>
>>335832308
everyone knows drivers may as well be like throwing darts with a blindfold on

there was an article going around a while ago about how pointlessly complex drivers were becoming thanks to directX and openGL

basically its a miracle they dont completely break game compatibility with every new driver update
>>
>>335832308
isn't this "downgrade" because they put finer MSAA stuff ? i heard amd had downgrade some shader to gain perf in a driver maybe they just making their shader and filter finer ? would that be possible
ps: i don't know shit about filter shader et antialisasing stuff
>>
Here on Linux I'm using nouveau with the experimental BOOST patches not yet in mainline Linux.

I'm getting really decent performance with the card now, compared to what I was getting before. The only issue is it causes my system to crash sometimes now.

>>335835725
This could possibly just be an accidental regression, not an intentional thing.

It's not uncommon for developers to release a newer version of software that actually introduces new bugs the older versions didn't have. It's called a regression.

I'm not doubting Nvidia did this on purpose though.
>>
>>335833742
AMD drivers are pretty shit though. With anything later than 16.2.1 shitloads of games simply freeze on my 290X CF on W8.1 (they're fine on W10 though). FreeSync doesn't work and is fiddly as fuck sometimes. FRTC is absolute crap, RTSS does a much better job but FS doesn't work with it anymore (it used to in older drivers).
Thread replies: 55
Thread images: 7

banner
banner
[Boards: 3 / a / aco / adv / an / asp / b / biz / c / cgl / ck / cm / co / d / diy / e / fa / fit / g / gd / gif / h / hc / his / hm / hr / i / ic / int / jp / k / lgbt / lit / m / mlp / mu / n / news / o / out / p / po / pol / qa / r / r9k / s / s4s / sci / soc / sp / t / tg / toy / trash / trv / tv / u / v / vg / vp / vr / w / wg / wsg / wsr / x / y] [Home]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.
If a post contains personal/copyrighted/illegal content you can contact me at [email protected] with that post and thread number and it will be removed as soon as possible.
DMCA Content Takedown via dmca.com
All images are hosted on imgur.com, send takedown notices to them.
This is a 4chan archive - all of the content originated from them. If you need IP information for a Poster - you need to contact them. This website shows only archived content.