Titan X + 970 GTX Physx Benchmarks

[gerardfraser]

EDIT:All Benchmarks Run were 5 minutes in game with max settings in game.Single Titan X and Titan X + 970 GTX ran with high Physx.
Change graph,sorry for any confusion.
Just to clarify it is extremely hard to get every kill/blowing up stuff ETC with 5 minute frap runs in game exactly the same each time for Titan X and Titan X + 970 GTX .YMMV.

Neither setup was able to run AC4Blackflag/Metro 2033 Redux/Metro Last Light Redux without 100% stutter free.


I have 1920 x 1080 144Hz screen ,hence the 1920 x 1080 Physx benchmarks.

May be of interest to some.




Data table for easier reading





For anyone interested Frametimes for all benches with fraps viewer.
http://www.mediafire.com/download/g9xhyq7x9ladt24/Titan+X+Frametimes+With+Viewer.7z

 

[Gripen90]

Good test ! Not really a difference (Titan X is strong enough to handle physx on it's own). Minimums FPS have improved in some games with the 970 as PhysX, but then again minimum FPS should be something you experience of because then it will lower your average FPS a lot.

 

[VincentHanna]

So just to clarify, the Titan X benchmarks on it's own, didn't have PhysX active?

 

[africanos23]

So just to clarify, the Titan X benchmarks on it's own, didn't have PhysX active?

The OPs title mentions the Titan X is the primary card with the GTX970 running Physx in the games.

Titan X + GTX970 running for all games.

 

[VincentHanna]

Yes but there's two benchmarks on his graph for each game and only the one + the GTX 970 seems to have Physx on?

 

[Ste]

I assume this is physx on in all tests and we're looking at titan alone vs titan with the 970 taking the physx load away?

Interesting comparison

 

[africanos23]

As Physx is offloaded to the CPU if using only the Titan X, maybe OP should specify CPU and OC ?

 

[VincentHanna]

See, I read that as Physx off completely with just the Titan X?

Otherwise why state 'PhysX high' in one benchmark but not the other?

 

[Yaayuh!]

OP, please clarify whether PhysX is being used at all without the GTX970 and whether it's the Titan X using it.

One would assume it's the same settings in the comparisons (it has to be for it to be worth anything), just in one it's the Titan X using PhysX and the other it's the 970.

 

[kedge]

Interesting results, it seems that alice madness and batman arkham origins benefited the most with the 970 as the dedicated physx card, i can see a 20-30 fps difference in mins and max, although the titanx is no slouch on its own with physx enabled, but interesting to see the titanx drop 26 fps at 1080p

 

[gerardfraser]

Opps ,sorry for any confusion.Changed it to hopefully read all test were maxed out.

Also Titan X is still not overkill for 1080 screens.

 

[Yaayuh!]

Some good results there! The minimums are quite a lot higher, that's what i found with Borderlands 2 when using a dedicated PhysX card.

 

[Drojman]

Its weird on my Titan X a lot of Physx seems to crash for me. Borderlands 2 on anything other than low crashes right after its finished loading my character in that blue eyes warpy thing.

 

[FMTopfan]

We need more PhysX benchmarks like these here, good work

Some good results there! The minimums are quite a lot higher, that's what i found with Borderlands 2 when using a dedicated PhysX card.

Borderlands 2 and the TPS is well known to have broken PhysX. Get about 8 or 10 enemies fighting on screen and its a stutter feast at High with dedi PhysX, reducing it to Medium still looks and feels laggy even though FPS was 55 avg from glances at AB. TPS is worse because it features the highest PhysX with Ultra but same poor performance as Borderlands 2 as it uses the same modified UE3.

From the years of tweaking the config on and off, then reading about other peoples tweaks I'll never get great FPS on a single GPU with most options enabled. Closest I've seen to good performance was on the Nvidia forum where a couple of Titan's were used in SLI and the third as dedi PhysX.

Be interesting if someone kind enough like Greg *cough* has time to test a couple 980/TX in SLI and a half decent GPU for dedi PhysX??

PhysX creates more demand rendering extra effects on the primary card/s and the HUGE performance impact makes this a no go. Unless you sacrifice a lot in terms of texture/decals/game details options. I end up using Low and that's equivalent to GPU PhysX off along with the effects - may as well have bought an AMD card or maybe 600-series is bad for dedi PhysX . Btw, I'm using a 680 for PhysX and 780 Ti primary card and Max Perf used with latest drivers.