GTX 980 SLI + GTX 750ti for Physx benchmarks

L

lucidrenegade

Limp Gawd
Joined
Nov 3, 2011
Messages
388
Just picked up two EVGA GTX 980 Superclocks and a EVGA 750ti Superclock for Physx. Figured I'd throw out a couple of benchmarks for anyone that's interested. Note that these are at 3840x2400 (1920x1200 monitor with DSR 4.00x). I'm waiting on my 4k monitor to ship.

Batman: Arkham Origins - All settings maxed, Vsync off in game and set to adaptive in Nvidia control panel
Single 980 + 750ti Physx dedicated - min 24, max 42, avg 31
Single 980 + 980 Physx dedicated - min 26, max 42, avg 32
2 x 980 in SLI + 750ti Physx dedicated - min 35, max 60,avg 54
2 x 980 in SLI, Physx set to Auto Select - min 30, max 59, avg 45

Having a dedicated Physx card makes the game smoother. Running Physx on one of the SLI cards causes perceptible micro-stutter.

I can run a few benchmarks in other games if anyone is interested. What's the best tool for checking FPS for a game without a built in benchmark? FRAPS?
 
Well clearly you should of had tri 980s with one dedicated to Physx! (Joking)

Thanks for the benches. I love data. When I looked into a dedicated physx card that's the gist I got - especially helps smoothness and minimums. When I looked at the % increase it was better than the 980 value curve so I snagged one.

I just got the same game for my daughter for Christmas for $7 off steam & a 360 controller. She's going to be thrilled. She loves the cat.
 
Quick question regarding a dedicated card for physx, i have sli 980s, when you install a different series card do you just install the drivers for that card and then you have 2 sets of drivers running at the same time?
 
Marcdaddy said:
Quick question regarding a dedicated card for physx, i have sli 980s, when you install a different series card do you just install the drivers for that card and then you have 2 sets of drivers running at the same time?
Click to expand...
GTX 750 Ti uses exactly the same driver package as the GTX 980, so there are no issues whatsoever.
 
Do you have Elder Scrolls Online?

I'd like to see how that goes. Single 980, plus the rest you did.

As soon as a 980 with a larger then 4gb frame buffer hits I plan to pick one up. Might use my 680 for a 4th monitor and physics.
 
Can you run a bench with just 1 x 980, and then the 1 x 980 + the 750ti for Physx?

I have a single 980 and was debating about heading down to the shop and picking up a cheap-o 750ti for Physx. I don't want to spend the cash on a 2nd 980 and go SLI, but was curious if having a 750ti for physx would help in heavy scenes of Far Cry 4 and maybe ARMA 3 (3440 x 1440)

Thx if you can test 1 980 vs 1 980 / 750ti and if you "feel" a difference in a heavy game.

(edit; not sure if Far Cry 4 is even a Physx game, it's not on the list on nvidia.com)
 
Last edited:
skypine27 said:
Can you run a bench with just 1 x 980, and then the 1 x 980 + the 750ti for Physx?

I have a single 980 and was debating about heading down to the shop and picking up a cheap-o 750ti for Physx. I don't want to spend the cash on a 2nd 980 and go SLI, but was curious if having a 750ti for physx would help in heavy scenes of Far Cry 4 and maybe ARMA 3 (3440 x 1440)

Thx if you can test 1 980 vs 1 980 / 750ti and if you "feel" a difference in a heavy game.

(edit; not sure if Far Cry 4 is even a Physx game, it's not on the list on nvidia.com)
Click to expand...

Farcry 4 uses Gameworks but not the Physx portion. Games I play/will be playing is Batman, Borderlands, Metro LL and the Witcher all make good use of it. Nvidia is still adding new technologies to Physx too, I think we'll continue to see it used.

I can bench Batman Arkham City later with only a 980 or 980/750ti if no one else gets to it.
 
Thx for the info.

I see now that my 2 x heaviest games, Far Cry 4 and ARMA 3 do not use Nvidia PhysX. I might still go pick up a 750ti anyway just because they are so cheap and don't generate any heat or noise.

Like you said, maybe in the future more titles will use hardware PhysX.

Thx for any more benchmarks....

(Shadows of Mordor?)
 
Can I use gtx 670 as dedicated physixs card or same gen as gtx 970s? I cant fit a third graphics card until i upgrade motherboard and case. Well maybe just case
 
This is the review I liked. A 670 should help looking at the chart in the review below (has a 650ti with a 780). You could always try with and without it as dedicated Physx if you already have it. It's easy to switch in the nVidia control panel.

http://alienbabeltech.com/main/using-maxwells-gtx-750-ti-dedicated-physx-card/3/


I did some benches...
Batman Arkham City (4k DSR, FXAA, maxxed):

980 Only: Min 41 Avg 55 Max 67
980 w/ 750ti: Min 51 Avg 62 Max 70

The dedicated card especially seems to avoid stutter on the ice part of the benchmark. I justified my purchase based on the minimums. I paid $13.40 / frame for the 980 solo. The 750ti was a similar, sometimes much better, value curve. Then sometimes non-existent with unsupported games! :) Lucky for me I play a lot of Physx games.
 
Last edited:
Unknown-One said:
I'm curious if anyone has compared the 750 vs. the 750 Ti as a dedicated PhysX card.

Reason being, there are nice passive versions of the 750 available. A PhysX card that adds no additional noise makes for a much more tempting buy.
Click to expand...

It'd probably do well. I read a review somewhere that the author thought clock speed was really important. Since the passive card is clocked the same as the 750ti... Should work well. Had something to do with getting the calcuations back to the primary card quickly.

I only get 11% usage on my 750ti for batman. But even maxing out physx (I think the bench is called fluid mark) it only gets up to about 30-40%. I think it maxes out some portion of the card though where the usage isn't properly taken... Someone once said physx only takes 16 cuda cores, it may, but it consumes something major someplace else or I wouldn't be seeing 10 FPS boosts (~20%) increases in frames with a dedicated card.

It's really hard to find dedicated card physx reviews. I looked. A few months ago and there might be around 3.
 
Last edited:
How about a bench with the 750ti as the main and the 980 as dedicated Physx vs 750ti? Silly I know, but science.
 
Wine said:
How about a bench with the 750ti as the main and the 980 as dedicated Physx vs 750ti? Silly I know, but science.
Click to expand...

I'd try it because of curiousity but I don't know how to do this without moving cards.
 
I wasted some money this morning. I mean, really wasted :)

all run at 3440 x 1440, v -sync off, high/ultra settings.

Shadows of Mordor GTX 980:
Max 104
Avg 55
Min 33
GTX 980 + 750ti Phys-X
Max 247 (had to be a glitch)
Avg 55
Min 32

Batman AC GTX 980
Max 107
Avg 76
Min 34
Batman AC GTX 980 +750ti
Max 113
Avg 84
Min 37

ARMA 3 GTX 980, Altis beta benchmark available on the net
Avg 33 (doesn't feel right, game play is smooth. Bench mark might be quite outdated)
ARMA 3 GTX 980 + 750Ti
Avg 29 (see note above)

Bioshock Infinite GTX 980
Max 385 (glitch)
Min 8
Avg 82
Bioshock Infinite GTX 980 + 750ti
Max 248
Min 11
Avg 82

Company of Heroes 2 GTX 980
Max 58
Min 17
Avg 37
w/ 980 + 750Ti
Max 58
Min 19
Avg 37

Basically if you unless your playing Batman A/C... Then DONT waste your money :)
 
skypine27 said:
I wasted some money this morning. I mean, really wasted :)

all run at 3440 x 1440, v -sync off, high/ultra settings.

Shadows of Mordor GTX 980:
Max 104
Avg 55
Min 33
GTX 980 + 750ti Phys-X
Max 247 (had to be a glitch)
Avg 55
Min 32

Batman AC GTX 980
Max 107
Avg 76
Min 34
Batman AC GTX 980 +750ti
Max 113
Avg 84
Min 37

ARMA 3 GTX 980, Altis beta benchmark available on the net
Avg 33 (doesn't feel right, game play is smooth. Bench mark might be quite outdated)
ARMA 3 GTX 980 + 750Ti
Avg 29 (see note above)

Bioshock Infinite GTX 980
Max 385 (glitch)
Min 8
Avg 82
Bioshock Infinite GTX 980 + 750ti
Max 248
Min 11
Avg 82

Company of Heroes 2 GTX 980
Max 58
Min 17
Avg 37
w/ 980 + 750Ti
Max 58
Min 19
Avg 37

Basically if you unless your playing Batman A/C... Then DONT waste your money :)
Click to expand...

I think the only game you tested that uses GPU Physx was Batman.
 
skypine27 said:
Basically if you unless your playing Batman A/C... Then DONT waste your money :)
Click to expand...
Why did you test a PhysX card in a bunch of games that don't evnen support hardware PhysX?

That's akin to testing a sound card with a silent movie, and then proclaiming the card is worthless for anything :rolleyes:
 
Unknown-One said:
Why did you test a PhysX card in a bunch of games that don't evnen support hardware PhysX?

That's akin to testing a sound card with a silent movie, and then proclaiming the card is worthless for anything :rolleyes:
Click to expand...


Because unless you plan to play the borderlands and batman series for the rest of your life, these are going to be the results you are getting.

You can keep hoping Physx is going to take off, but it didn't back in 2008, it didn't when nvidia bought them and its been used just enough every year to keep it from completly dieing.

As you can see in the results, don't waste your money on a dedicated card, a 980 is plenty for the 2 or 3 frames you are going to get with a dedicated
 
OmgitsSexyChase said:
Because unless you plan to play the borderlands and batman series for the rest of your life, these are going to be the results you are getting.
Click to expand...
We're telling you it's disingenuous test a PhysX card in a game that DOES NOT USE PHYSX, and then claim PhysX cards don't do jack. We all already know that a PhysX card will have no impact on performance in a game that does not use hardware PhysX, so running benchmarks in said games is beyond pointless.

That's like installing a high-end gaming card and then testing it with game that uses 100% software-accelerated rendering... then proclaiming all gaming cards don't do jack because there was no noticeable performance increase. No shit, an invalid test leads to invalid results.

OmgitsSexyChase said:
You can keep hoping Physx is going to take off, but it didn't back in 2008, it didn't when nvidia bought them and its been used just enough every year to keep it from completly dieing.
Click to expand...
You assume wrongly. I'm not hoping for PhysX to take off. I really don't give a shit how many games use it.

What I care about is being able to run my games with all the bells-and-whistles enabled. Sometimes, that might require a dedicated PhysX card. If that's what a particular games needs, that's what it needs *shrug*
 
Last edited:
OmgitsSexyChase said:
Because unless you plan to play the borderlands and batman series for the rest of your life, these are going to be the results you are getting.

As you can see in the results, don't waste your money on a dedicated card, a 980 is plenty for the 2 or 3 frames you are going to get with a dedicated
Click to expand...

There are actually quite a few titles that use PhysX outside of Batman and Borderlands and I'm sure there will be more titles that use it in such a way that a dedicated card will add considerable performance. Whether or not that matters to you or not is, of course, another matter entirely! ;)

I think the point of this thread was to demonstrate whether or not a discrete PhysX solution is of value where applicable, considering it sure as hell is useless to have a card dedicated to a task that most games don't use ;)

I've toyed with dedicated PhysX stuff off and on several times starting with a GTX275+8800GT and, in any game that can use a dedicated card, the gains on minimum framerate are usually well worth the addition if you can do it mostly for free. The only reason I have a dedicated card at the moment is because I can, and it's been pretty sweet for the games that have all the nifty particle effects and such :D

In my experience you can generally expect to gain about 5-10fps on the minimum FPS and as much as 30fps on the max FPS depending on the primary+secondary card config (I'm assuming no SLI or X-fire) and the platform you're on. That's probably not worth spending $110 on a GTX750 for no other reason than PhysX-capable titles, but if you have something laying around to do it, and/or you can score a card for cheap then it's fun to have :cool:
 
All of my planned future games use Physx... Batman Origins, Batman Arkham City, Star Citizen, The Witcher 3, Borderlands. So it makes sense to me..... Oh and Starcraft II Legacy of the Void which doesn't use Physx. So one title I am going to be buying in 2014/2015 doesn't use it.

$100 for a Physx card is nothing and if it lets me lock 60 FPS with good AA it's totally worth it.

I would love to see more data on actual Physx games.
 
Last edited:
Im glad to see Star Citizen using it (if that game is ever actually finished), though I wish Elite Dangerous and Far Cry 4 had jumped on it too.

Like the other poster said, Physx has been around since 2008 and never really took off.
 
skypine27 said:
Im glad to see Star Citizen using it (if that game is ever actually finished), though I wish Elite Dangerous and Far Cry 4 had jumped on it too.

Like the other poster said, Physx has been around since 2008 and never really took off.
Click to expand...

Yeah I was a bit bummed FC4 didn't use it at least for the fur. It could of been so much better. Way better than the hideous animals in FC3 though.

I justified my purchase off of current games - anything in the future is a bonus.

nVidia is still dumping money into it which means we'll continue to see it. Because.. fruit loops in milk...
https://www.youtube.com/watch?v=1o0Nuq71gI4&list=UUHuiy8bXnmK5nisYHUd1J5g
 
For comparison purposes, currently playing batman arkham city.

Gtx 690 @ 1202 MHz vsync off
All setting maxed , 32xcsaa + sweetfx smaa + dsr 4434x2494 (1440p)

Benchmark

Min- 38
Max- 88
Ave- 65
 
I did something similar a while back (about 2 years ago) for Borderlands 2 with a GTX 680 as primary and then testing a GTX 580 SC and a GT 640 as dedicated PhysX:
http://1pcent.com/?p=135

Similar results. Although now that I have a G-Sync monitor - I'd recommend putting your money that way to get the "smoothness" in every game, and not just PhysX-enabled ones. :)
 
You must log in or register to reply here.
Back
Top