IS NVIDIA RUINING YOUR PERFORMANCE?

[jakkuh_t]

Rumors of NVIDIA's planned obsolescence have been around forever but now we finally investigate.

 

Buy NVIDIA Graphics Cards: 
Amazon: http://geni.us/FynNMSn
Newegg: http://geni.us/JhScB

 

 

[Imbellis]

nah.

[ProjectBox153]

I still have a card that barely gets 50 fps. In MINECRAFT. With low settings. I don't think performance is my top priority. 

[Vandorlot]

3 minutes ago, Jamiec1130 said:

I still have a card that barely gets 50 fps. In MINECRAFT. With low settings. I don't think performance is my top priority. 

Minecraft uses mostly cpu and not gpu.

[TheGleaner]

Why all caps though?

[DildorTheDecent]

2 minutes ago, TheGleaner said:

Why all caps though?

THAT'S HOW YOUTUBE WORKS NOW MY DUDE. IF IT ISN'T ALL CAPS THEN IN THE BIN IT GOES

[TheGleaner]

1 minute ago, DildorTheDecent said:

THAT'S HOW YOUTUBE WORKS NOW MY DUDE. IF IT ISN'T ALL CAPS THEN IN THE BIN IT GOES

I just picture linus yelling this on the street, and people looking at him like he's crazy...

 

Edit: so no wonder why tq and csf isn't as popular.

[Falconevo]

Why is the video unlisted :o?

[IanLuminus]

Here's something specific that I was able to replicate - GTX670 stock speeds with almost any game that does mid-range usage on one monitor and even something as simple as a 720p Youtube video on the other monitor. 38x.xx drivers will cause stuttering. Not sure where it started, but I know for sure the 361.75 driver works wonderfully. I can confirm this is not a hardware-specific issue because a friend with a GTX660 and different CPU/RAM/etc experiences the same problem.

[jakkuh_t]

8 minutes ago, Falconevo said:

Why is the video unlisted :o?

While we are getting the video description and stuff ready, the forum peeps get a litte "early access".

[Majestic]

This....is actually a good video @LinusTech. Kudo's. It needed to be said, now unlist it dude.

[Top_Hat_Tomato]

27 minutes ago, Jamiec1130 said:

I still have a card that barely gets 50 fps. In MINECRAFT. With low settings. I don't think performance is my top priority. 

I think it's something wrong with your computer, I used to run minecraft at around 60fps with a 8200m (which is motherboard integrated graphics) and ran at 400mhz.

[mrchow19910319]

From now on this is gonna be my wallpaper: 

 

[SanityAgathion]

1 hour ago, IanLuminus said:

Here's something specific that I was able to replicate - GTX670 stock speeds with almost any game that does mid-range usage on one monitor and even something as simple as a 720p Youtube video on the other monitor. 38x.xx drivers will cause stuttering. Not sure where it started, but I know for sure the 361.75 driver works wonderfully. I can confirm this is not a hardware-specific issue because a friend with a GTX660 and different CPU/RAM/etc experiences the same problem.

It may not had been deliberate though, more like driver development team shifting their priorities somewhere else. Does this go for all of 38x.xx driver line, or did it get fixed with another driver update? In which case, if NVidia really broke the performance for example by requiring cards to have newer features in hardware, they should recommend older working drivers for download rather than new. Which also means there won't be optimizations for new games and ... yeah, obsoleting the card.

[mrchow19910319]

I clearly remembered my GTX780's performance tanked a bit after the gtx970 launch. 

In the same game. (battlefield 4) 

[IanLuminus]

2 minutes ago, SanityAgathion said:

It may not had been deliberate though, more like driver development team shifting their priorities somewhere else. Does this go for all of 38x.xx driver line, or did it get fixed with another driver update? In which case, if NVidia really broke the performance for example by requiring cards to have newer features in hardware, they should recommend older working drivers for download rather than new.

I tried about four different driver sets before going back to the 361.75, all in the late 37x.xx-mid38x.xx line. At first, I attributed it to a Windows 10 thing. Then saw it continue to happen when I reinstalled 7. From what I can find on the googles, this is an issue affecting all pre-Pascel cards.

 

Small note, it was NOT due to the GPU hitting 100% usage.. was happening even under the 915MHz clock (700-750MHz autoclock) and 40-50% usage.

[cesrai]

1 hour ago, mrchow19910319 said:

I clearly remembered my GTX780's performance tanked a bit after the gtx970 launch. 

In the same game. (battlefield 4) 

Please do benchmarks with both drivers and then share your results.

[mrchow19910319]

Just now, cesrai said:

Please do benchmarks with both drivers and then share your results.

nah... i dont own a 780 right now and not even a gaming desktop/

sold those long time ago.sorry

[masterg7]

Personally i think this is true on both sides, i have a XFX R9 380 4gb. I got overwatch roughly around christmas time, i was able to play 1080p epic settings with temps sitting around 78-84c and running around 100fps, here in the past 3 months ish, one of the driver updates that were supposed to improve performance actually reduced my performance due to thermal throttling. i've had to step my settings back to ultra just to keep it from hitting 100c, the game settles around 96-98c at all times without stuttering. when it hit 100c it stuttered real bad. Ive also had to step back from the 17.7.2 driver because it was crashing during games and anytime the screen sent to sleep just sitting on the desktop. so 17.7.1 saves the day till they fix the 17.7.2.

[crystal6tak]

Managed to dig up an article I read over a year ago

 

They compared the 780 ti, 290x, and Titan using drivers from 2013 and 2016. I feel like this shows the full picture much better than what Linus offered:

 

I did some calculations, I disregarded the Secret World and Metro Last Light as they were oddballs. Secret world didn't work on the 2016 AMD drivers, and AMD specifically optimized Metro Last light's benchmark tool (actual ingame performance was not improved that much).

 

Here's my numbers:

From 2013 to 2016:

780 Ti improved 2.9%

290X improved 7.2%

 

Relatively:

In 2013, 780 ti was 12.7% better than 290x

In 2016, 780 ti was 8.4% better than 290x

 

Personally, I don't think this is a big deal. Feels like a win-win to both 780 Ti and 290X users. If you bought a 780 Ti at a premium over the 290X, you can still be happy that your card is outperforming the 290X. if you bought a 290X, you're getting a bit more out of your card, but not enough to beat the 780 ti. 

 

And really, is a few percent performance difference that important? I rather them improve the stability of the drivers, or even offer more neat features (such like shadowplay).

 

Just buy whatever suits you best now, and enjoy your card!

[Princess Luna]

The only problem is that lots and lots of people allow Geforce Experience to update the driver and that is the complication, it leaves tons of left overs that break the performance, if every one would DDU in safe mode before updating to the latest driver there wouldn't any issues.

[AshleyAshes]

16 minutes ago, Princess Cadence said:

The only problem is that lots and lots of people allow Geforce Experience to update the driver and that is the complication, it leaves tons of left overs that break the performance, if every one would DDU in safe mode before updating to the latest driver there wouldn't any issues.

This is perhaps another claim that should be investigated.  I'm sure that people will give lots of anecdotal stories but stuff like this is about doing the opposite, doing concrete testing to see if something is happening or if it's the subjective experience of the user.

 

To me, the idea seems silly.  Files are used or they are not used.  The idea that unused files have an influence on used files and the performance of the data in those files is ridiculous but I'd be interested in seeing objective testing done on the concept.  It'd be easy enough replicate.  Install a driver, reboot, bench, install a newer driver, reboot, bench, and so on.  Stacking driver installations atop each other the whole way.  Then get a clean installation for both the oldest and newest driver versions used and compare the two sets of data.

[mech]

On 8/16/2017 at 1:35 PM, Jamiec1130 said:

I still have a card that barely gets 50 fps. In MINECRAFT. With low settings. I don't think performance is my top priority. 

I topped you I get 4 fps in Minecraft 

[Morgan MLGman]

On 8/17/2017 at 8:16 AM, crystal6tak said:

SNIP

The problem is that you chose a chart with only old games in it, here's one with newer ones (a.k.a. more relevant one):

That is when Maxwell came out and Nvidia stopped giving a damn about Kepler and you can clearly see the pattern. They stopped optimizing for new games and the performance is bad compared to competing cards from the same era.

[ProjectBox153]

On 8/21/2017 at 6:43 PM, ScrappyZeDog said:

I topped you I get 4 fps in Minecraft 

I didn't mention my main laptop. It gets 1.