Nvidia's New Driver Tested & Benchmarked, Comes Up Short on Expectations.

[TERAFLOP]

Anandtech has released its benchmarking results for Nvidia's new driver that's supposed to go head to head with AMD's Mantle API.
And here are the results.

http://www.anandtech.com/show/7926/nvidia-releases-33750-beta-driver-offers-significant-performance-improvements

 

Notable gains in Bioshock & Metro up to 11%

 

Notable gains in Bioshock, Metro & Grid 2 up to 9%. The jump in Rome II is due to the R331 drivers lacking SLI support for Rome which was introduced in the latest R337 drivers.


These gains come up considerably short on Nvidia's previous claims.

 

[Mitrailleuse]

bad benchmark,

this driver is meant to reduce cpu bottlenecks,

they should have tested with weaker cpu's.

[Katness Everdeen]

I immediately sensed a red flag when Nvidia started comparing its flagship 700$ 780 Ti to AMD's 550$ card the R9 290X which instead of its natural competitor the 500$ GTX 780.

[Flight1sim]

UGGHH i still cannot use shadowplay for laptop...
HALP plz

EDIT: YEAH this is for Geforce Experience 2 not the drivers

[CarlosRex]

bad benchmark,

this driver is meant to reduce cpu bottlenecks,

they should have tested with weaker cpu's.

Huh funny. Nvidia doesn't seem to think so since they "showcased" the performance gains using a one thousand dollar CPU which is the i7 3960X.

[Altecice]

TOLD YOU THIS WOULD HAPPEN!

[MetroP]

Just like every other major company i.e. Apple, Samsung. Nvidia will manipulate people with marketing and get away with it.

[remnbig31]

Maybe its for people that never had a driver installed in their life, maybe Nvidia started to drink the same CoolAid as AMD.....who knows

[EmmaLong]

Nvidia "We can make DX11 faster than Mantle"
Mantle "Yeah... right"
DX11 "Uh... guys...what are you talking about?"
Nvidia "Keep your mouth shut DX11 and do as we say"
DX11 "Uhm... OK" DX11 'Pees pants'
 

[Askew]

Wasn't the whole mechanism of the increased performance based on reducing CPU overhead?

[KieranFretwell]

Elric got some boost with his..

[One Who Craves Souls]

Some people claim it's not that good, others claim it is...  :unsure:

[kuddlesworth9419]

This is a really bad benchmark. They should be giving us the clear cut data. Eg. but the fps scores into each bar graph and compare them side by side. Saying 1.0 vs 1.07 means nothing.  Besides I have seen actual fps improvements on some benchmarks of over 10 fps in some games and I think there was one game that had over 30 fps boost. Star swarm gets a huge boost in performance for example. All the other games get slight performance boosts but still it's free performance that you didn't pay for so ther you go. Plus I am sure there are a bunch of games out there that will love this new driver. Arma 3 I bet will get a huge boost in performance.

 

http://www.tomshardware.com/news/nvidia-geforce-337.50-driver-benchmarks,26473.html

[Scia]

this.

[papeles]

guru3d sli benchmarks. 

 

http://www.guru3d.com/news_story/geforce_337_50_beta_sli_performance_benchmarks.html

[KieranBrown]

Huh funny. Nvidia doesn't seem to think so since they "showcased" the performance gains using a one thousand dollar CPU which is the i7 3960X.

they can't even get the CPU right, it's a 3960X lawl

[Xorbot]

As if we needed more of reason to keep hanging on to our Sandy Bridges, or hell.. even Broadwells.  I guess it's always a good thing to not have to spend money on CPUs.

[ACatWithThumbs]

guru3d sli benchmarks. 

 

http://www.guru3d.com/news_story/geforce_337_50_beta_sli_performance_benchmarks.html

Those increases are massive 27Fps more in Crysis 3

Also we need some benchmarks where they show us the minimum FPS because that's where the biggest change should be.

And we need different CPU set ups.

 

[Si3Rra_7]

Bullcrap , as usual . Instead of focusing money on DX12 and getting actually competitive gpus on the market ...

[ohJey]

This is a really bad benchmark. They should be giving us the clear cut data. Eg. but the fps scores into each bar graph and compare them side by side. Saying 1.0 vs 1.07 means nothing.  Besides I have seen actual fps improvements on some benchmarks of over 10 fps in some games and I think there was one game that had over 30 fps boost. Star swarm gets a huge boost in performance for example. All the other games get slight performance boosts but still it's free performance that you didn't pay for so ther you go. Plus I am sure there are a bunch of games out there that will love this new driver. Arma 3 I bet will get a huge boost in performance.

 

http://www.tomshardware.com/news/nvidia-geforce-337.50-driver-benchmarks,26473.html

You can't read a basic graph?  1.00 to 1.07 = 7% gain.  

[ACatWithThumbs]

You can't read a basic graph?  1.00 to 1.07 = 7% gain.  

The problem is that you can't benchmark something like this with that method.

We need to know the minimum FPS and we need different CPU configurations.

Just like with Mantle.

With this method you will only see a change if the game has a massive CPU bottleneck.

From what I've tested tons of games get a massive minimum FPS boost from 10Fps+. 

[bob345]

Huh funny. Nvidia doesn't seem to think so since they "showcased" the performance gains using a one thousand dollar CPU which is the i7 3960X.

 

it says 3960k in the pic you linked. just wanted to throw that out there

[DeViLzzz]

Well one good thing about Nvidia still whether these drivers are that good or not is that when you have their cards you can still play games the way they were meant to be played ... with physX. Oh so physexy!

[Notional]

it says 3960k in the pic you linked. just wanted to throw that out there

 Who knows what that is, it does not exist on intel's own homepage. It's either a 3960X or 3930K. 3960X is already unlocked which the k suffix usually stands for. The 3960x processor is like a 1000$ CPU.

[Advent]

This benchmark is horrible. I have seen nice FPS increase across quite a few titles with the recent driver update. I got 15+ FPS on BF4, ~20+ on C3, around the same on AC3 and even some improvements in Rust.