337.50 Beta Driver Benchmark results

[pabloottawa]

Not much of an improvement although I have not tested it "in the field"

 

[Flibolito]

Could you disable Hyperthreading and lower the clocks to say 2.4GHz? You CPU is not even close to a bottleneck here (which is great). I think the benefits here are in CPU limited scenarios, games like SC2 and other cpu bound games will show a huge difference me thinks.

 

[KainXS]

want to see hyperthreading disabled also

 

[Shou Miko]

i guess the Valley Benchmark don't see HT since it only write x6 at the end of the processor

 

[MxPhenom 216]

Benchmarks like this sre not going to see much of an improvement, since they are far from CPU limited.

 

[Slizzo]

I saw a pretty good improvement in Battlefield 4. My experience is a lot smoother now (when the server isn't rubber banding) on the newer beta versus the previous official driver.

 

[pabloottawa]

I also play BF4 and I did notice it was smoother..... I think Nvidia was making claims as outlandish as AMD about their mantle.... I have yet to see a 40% improvement though.

 

[Hilux SSRG]

I saw a pretty good improvement in Battlefield 4. My experience is a lot smoother now (when the server isn't rubber banding) on the newer beta versus the previous official driver.

Exactly for me as well.

 

[kn00tcn]

outlandish claims? where are the total war tests? how about sleeping dogs? if their claims are cherry picked, you have to run the same cherries

 

[TheHunter]

Why dont you try some cpu intense benchmarks instead?

Because comparing it to gpu bound benchmarks its pretty much useless.



here have a look at my boost scenario, I already posted this in another thread though

I got boosts in a lot of games, although I had to test with a little lower reso. by some to see that cpu bottleneck.




Batman AC

Physx - high; rest highest dx11, with fxaa - high @ 1080p

some 331.xx driver

min 43fps
max 113fps
avg 81fps


now 337.50

min 50fps
max 122fps
avg 86fps


----------------------------------------------------------------------
Hitman Absolution

tested lower reso 720p, level detail - ultra, no aa..


335.23
min 60-62fps (be it 4.5 or 4.7ghz)


337.50
min 82fps



a whooping 18-20fps boost by min fps

---------------------------------------------------------------------


JustCasue2

Desert Sunrise

335.23


337.50


Gained ~ 5-6fps on avg.

-----------------------------------------------------------------------


HardReset

ingame max, fxaa @ 1080p

335.23
min: 76-77fps (best of the best so far)

[Imgur](http://i.imgur.com/u2nuauG.jpg)

337.50
min: 80.5fps

[Imgur](http://i.imgur.com/fKcPSFP.jpg)

But then again, I never see this min fps.
min fps usually happened by that first big car explosion, now I saw ~ 92fps there and gpu usage to the max.. Idk what was with 335.23 though, but i remember some earlier drivers had ~ 80-90% gpu usage and min fps ~75fps..

----------------------------------------------------------------------

LostPlanet Colonies, 720p, dx10 rest highest, no aa. Tested for cpu bottleneck

335.23 or was it 331.87
min fps: 181fps

[Imgur](http://i.imgur.com/EE9YjsP)


337.50
min fps: 197fps

[Imgur](http://i.imgur.com/DDv4gEu)


cpu bottleneck:

[Imgur](http://i.imgur.com/0iUaQLs)

now min min 143fps, before with older ~132fps

------------------------------------------------------------------------


Batman AO, physx - high, 1080p, fxaa, dof - normal

335.23
min fps: 53

[Imgur](http://i.imgur.com/7jhsCMn.jpg)

337.50
min fps: 56

[Imgur](http://i.imgur.com/eGlQ45L.jpg)

-------------------------------------------------------------------------
Monster Hunter 1

335.23

[Imgur](http://i.imgur.com/Ah5D8F6.jpg)

337.50

[Imgur](http://i.imgur.com/pC61523.jpg)

-------------------------------------------------------------------------


3dmark06

Old score with 570gtx @930mhz, cpu @ 4.7ghz,

Driver 331.40
34480


http://i.imgur.com/MVwnGjD.png



Now 580gtx @ 900mhz, cpu @ 4.6ghz.

38702



and this cpu bound location, now min ~64fps, before ~58fps + before I also had a little higher cpu OC


--------------------------------------------------------------------------

RAGE - cuda transcode

335.23
385

[Imgur](http://i.imgur.com/T8VxdOq.jpg)

337.50
394 (after 4th try), 2 times it was ~ 380-386

[Imgur](http://i.imgur.com/kTgTxGo.jpg)

---------------------------------------------------------------------------

Sleeping Dogs, all max, fxaa;
(boost 1-2 fps by avg), was gpu bottlenecked all the time,

335.23

[Imgur](http://i.imgur.com/Zd1BHly.jpg)

337.50

[Imgur](http://i.imgur.com/LjRYcDw.jpg)


335.23 already fixed those cpu bottleneck for me..

----------------------------------------------------------------------------------------

StarSwarm - extreme follow @ 1080p

335.23
Average FPS: 37.46

337.50
Average FPS: 39.94
_____________________________________________________________________________________________________

The only game that had no affect, sadly -_- was RedFaction Armageddon (dx11), same fps drops by explosions and X3TC bench - not a single fps increase gpu usage ~30-40%.




Overall quite impressed.

 

[pabloottawa]

Guys,

I don't get off on benchmarking so if you all want these different tests done, do them yourself and share the info.

Cheers

 

[MxPhenom 216]

Guys,

I don't get off on benchmarking so if you all want these different tests done, do them yourself and share the info.

Cheers

What a pointless thread then. Make benchmark thread, puts one benchmark.

 

[Steevo]

So the point here is old Nvidia games anyway, Batman 2011...its old enough to be potty trained!!!

And now we are going to start benchmarking to show off our epeen by downclocking and choking it? Its amazing how Nvidia can give its users the Cleavland Steamer and they go 2 girls on it.

 

[Slizzo]

Guys,

I don't get off on benchmarking so if you all want these different tests done, do them yourself and share the info.

Cheers

You do know that this driver was tailored to only improve performance in those games that they listed right? And that performance can vary from system to system, so some will see improvement, some wont?

 

[Flibolito]

This is a great trend we are seeing. PC gaming in the future will be less expensive and PCs will last longer requiring only GPU upgrades once in a while. This kind of software technology will make PC game much more accessible for many people. Frankly as this matures I see anyone with a i5/i7 Sandy bridge and up cpu should be fine for years especially when overclocked.

 

[Fluffmeister]

Computer Base tested an 3960K with a combo of either two or four cores disabled along with HT thrown in, and in all cases they saw improvements over 335.23.

The biggest gains were seen when running the fewer cores/more CPU limited scenarios:

http://www.computerbase.de/2014-04/geforce-337.50-cpu-skalierung-benchmarks/

 

[MxPhenom 216]

You do know that this driver was tailored to only improve performance in those games that they listed right? And that performance can vary from system to system, so some will see improvement, some wont?

Wrong, the performance boost will happen across most DX9 and DX11 games, especially CPU limited, though GPU limited games are showing big boosts too. The ones listed are just the more substantial ones. Battlefield 4 has been getting some big boosts for people, but its not listed.

 

[pabloottawa]

What a pointless thread then. Make benchmark thread, puts one benchmark.

Yeah you're right. Because I just live to please everyone.. Do you have two hands and a PC???? Yes? then do it yourself. Ignored

 

[MxPhenom 216]

Yeah you're right. Because I just live to please everyone.. Do you have two hands and a PC???? Yes? then do it yourself.

I actually don't have two hands.......

 

[FX-GMC]

Yeah you're right. Because I just live to please everyone.. Do you have two hands and a PC???? Yes? then do it yourself. Ignored

Useless thread. Jerk OP. Thread should be closed.

 

[pabloottawa]

At least it was a contribution asshole, I don't see you posting benchmark results.

 

[FX-GMC]

At least it was a contribution asshole, I don't see you posting benchmark results.

I don't usually post benchmark results, but when I do they are actually useful.

Posting a Valley run for the new driver....not useful.

 

[Tatty_Two]

Methinks this thread has descended into....... erm...... nothing really, thanks to the OP for sharing his findings, before it turns into an all out holocaust I suggest it closes..... thank you!