Some Interesting Batman GPU Results

W

WabeWalker

Supreme [H]ardness
Joined
Sep 3, 2007
Messages
4,508
Batman Arkham Asylum has a benchmark utility which basically runs the engine and records your highs and your lows. I thought that this might be mildly interesting to some of you out there.

Batman for the PC was held back of course because of PhysX, and I noticed in the demo (and now in the game itself) that if you want to run the game with PhysX at the highest setting then Rocksteady recommends that you have two GPUs (doesn't have to be SLI) and that you allow one of the GPUs to completely work on PhysX.

I have a single GTX 275 (along with a core i7), and I was shocked at the benchmark result. This was with everything set to the highest possible setting, including PhysX, and with 4X anti-aliasing, at 1920x1080.

Min: 20 fps.
Average: 35 fps.
High: 60 fps.

My frame-rate, by the way, was capped at 60.

I had been expecting better results than that. But people at the official forum were saying that PhysX is a killer - and indeed when I turned off PhysX my numbers went through the roof.

The thing is, this summer my GTX 275 replaced a 9800gt, which has been sitting in the closet ever since - so I decided, why not install the 9800gt and use it for the PhysX. I did just that and ran the benchmark again. Here's what happened.

Min: 39
Avg: 60
High: 60

So it actually works. Dedicating one card to PhysX had a pretty substantial impact the on the performance. My minimum and average frames jumped significantly. Anyhow, I'm glad I kept hold of my 9800gt, because I had been planning on selling it for $40-$50. It'll at least be worth it for Batman, I think.

It's not a big deal or anything. I just thought I would post that.
 
And people called AGEIA crazy for making a dedicated PPU, saying that a GPU would be more than enough to run both graphics and physics...go wonder.
 
Looks like I'm gonna keep my 9800gt/9400gt setup for now.
(it accelerates Softimage 7.5, though I have to redo ALL my simulations now!)
 
Atech said:
And people called AGEIA crazy for making a dedicated PPU, saying that a GPU would be more than enough to run both graphics and physics...go wonder.
Click to expand...

AGEIA is crazy.

Why buy a dedicated PPU?
Before nVidia, there was no sane reason to even bother with GPU/PPU physics.

EDIT: Wierd thing, what in this world is Intel/AMD doing with Havok!?
 
jeremyshaw said:
AGEIA is crazy.

Why buy a dedicated PPU?
Before nVidia, there was no sane reason to even bother with GPU/PPU physics.
Click to expand...

Before 3Dfx there was no reason to bother with 3D...keep living in the past.

I have a nice place on my ignore for you, your FUD is boring.
 
Atech said:
Before 3Dfx there was no reason to bother with 3D...keep living in the past.

I have a nice place on my ignore for you, your FUD is boring.
Click to expand...

FUD?

EDIT: nevermind, it's the zealot again

EDIT2: Except GPU physx hasn't seen any games of note, while havok has.

EDIT3: GPU physx has found no place in gaming... if you need 2 cards to speed crap up, then why bother? (you are the one, after all, who decided to point out dual card solutions are 'flaky'.)

EDIT4: I'm numbering these so you don't whine about 'ninja edits'
 
Last edited:
Hmmmm. You're right. To get 59.5 you'd have to play only 42 seconds with a cap of 60fps to achieve avg of 60. It just seemed it should take longer. :(
 
WabeWalker said:
Yes, I did make a mistake.

I meant to write down 58 for my Average. Sorry about that.
Click to expand...
OP those results still look to good to be true. I have seen results for at least 3 people that have a gtx275 or gtx285 all using a 9800gt or better for physx and they still have minimums in the 20s during the benchmark with high physx. that just seems like way too big of a jump from adding 9800gt plus the fact that none of the other benchmarks I have seen with a similar setup have been close to that as for as minimum framerate goes. :confused:
 
I turned V-Sync off and ran the benchmark again. (This is the GTX 275/9800gt setup, with the 9800gt obviously handling PhysX).

The numbers were a bit different.

Min: 44
Avg: 75
High: 110

That's weird. My minimum was actually a bit higher with V-sync switched off? Also, my fans went ballistic. That didn't happen when V-sync was enabled.

Also, please note that I'm not saying that PhysX is the second coming or anything. I realize that very few games feature it prominently. I just thought that this might be worth noting.

One thing I can say is that I actually did notice a big difference at the beginning of the game when the 9800gt was handling the PhysX. This is the section of the game where The Joker is in the chair, and everyone's in that huge room, with all the flags flying above.

I could actually tell that the game was running more smoothly.

Before, when the GTX 275 was handling the PhysX by itself, that section of the game was just a little bit... I don't want to say stuttery... but somehow not quite smooth. When I returned to that part of the game, with the 9800gt enabled for PhysX, the game ran like butter. I mean, I could actually feel the difference. It was very noticeable.

Anyway, for whatever it's worth. There it is.
 
cannondale06 said:
OP those results still look to good to be true. I have seen results for at least 3 people that have a gtx275 or gtx285 all using a 9800gt or better for physx and they still have minimums in the 20s during the benchmark with high physx. that just seems like way too big of a jump from adding 9800gt plus the fact that none of the other benchmarks I have seen with a similar setup have been close to that as for as minimum framerate goes. :confused:
Click to expand...

I don't know what to say.

The numbers are exactly as they appeared on my screen.
 
WabeWalker said:
I turned V-Sync off and ran the benchmark again. (This is the GTX 275/9800gt setup, with the 9800gt obviously handling PhysX).

The numbers were a bit different.

Min: 44
Avg: 75
High: 110

That's weird. My minimum was actually a bit higher with V-sync switched off? Also, my fans went ballistic. That didn't happen when V-sync was enabled.

Also, please note that I'm not saying that PhysX is the second coming or anything. I realize that very few games feature it prominently. I just thought that this might be worth noting.

One thing I can say is that I actually did notice a big difference at the beginning of the game when the 9800gt was handling the PhysX. This is the section of the game where The Joker is in the chair, and everyone's in that huge room, with all the flags flying above.

I could actually tell that the game was running more smoothly.

Before, when the GTX 275 was handling the PhysX by itself, that section of the game was just a little bit... I don't want to say stuttery... but somehow not quite smooth. When I returned to that part of the game, with the 9800gt enabled for PhysX, the game ran like butter. I mean, I could actually feel the difference. It was very noticeable.

Anyway, for whatever it's worth. There it is.
Click to expand...
well yeah minimums can easily go up with vsync off. I just dont understand how your minimums are that high when nobody else is even close to that. I am sure you have seen some of the results on the official forums and nobody is getting in the 40s for minimums with those settings. not even people with gtx260 and gtx285 cards being used for dedicated physx cards are getting those minimums with physx on high. I am not knocking you but I just hate when things dont make sense. lol
 
That is impressive, and I did really like the added effects that turning on physx added to this game.
 
cannondale06 said:
well yeah minimums can easily go up with vsync off. I just dont understand how your minimums are that high when nobody else is even close to that. I am sure you have seen some of the results on the official forums and nobody is getting in the 40s for minimums with those settings. not even people with gtx260 and gtx285 cards being used for dedicated physx cards are getting those minimums with physx on high. I am not knocking you but I just hate when things dont make sense. lol
Click to expand...

It's good to be skeptical. I'm that way myself.

I can't explain why other people have posted low minimums even with a second GPU handling the PhysX. Surely they had the most recent Nvidia drivers installed? The latest driver actually adds increased PhysX support for Batman. Other than that, I can't fathom why they would still be getting low minimums? But really they shouldn't be. If I can get a low of 38 then others should be able to get that as well.

All I can say is that the demo ran beautifully for me on my GTX 275.

But then I started the game tonight, and in that big starting room, with all those flags flying above, there was something wrong... I could actually feel it... so I ran the benchmark, and was dismayed at seeing that my low was indeed 20 fps.

Then I installed my 9800gt (I didn't hesitate to do this, because I've been waiting for this game since I played the demo, and obviously I want it to run well) and I could tell right away, even without running the benchmark, that everything was different. You know how you get a silky smooth feeling when you're running a game that's working well... yeah, that's what it felt like.

No way am I removing my 9800gt now. I've noticed that the temps are a bit higher, and that in other games, my fans are coming on sooner than they normally would've (I have no idea why), but when I'm finished with Batman I'll probably take the card out.
 
The CPU type/speed will affect the minimum framerates.
ie i7 at high speed will give better results than any socket 775 chip.

Also some peoples results are from gameplay, not the canned benchmark.
Use Vsync off when making comparisons unless it has been specified that Vsync must be on.
 
Nenu said:
The CPU type/speed will affect the minimum framerates.
ie i7 at high speed will give better results than any socket 775 chip.

Also some peoples results are from gameplay, not the canned benchmark.
Use Vsync off when making comparisons unless it has been specified that Vsync must be on.
Click to expand...
I was going by the built in benchmark and some of the people I was referring too also had an i7 cpu. thats why I am perplexed by his numbers.
 
A newer/less polluted OS install may help.
the OS itself can make a difference, I have found I get a smoother game play in Windows 7 than XP for example.

There there are drivers and their configuration.
Some older drivers are faster than some newer drivers and vise versa.
Then there are Beta drivers, some of them have specific enhancements for new games.
Some people like to boast and turn down driver quality settings to get higher numbers...
 
I didn't check my actual frame rates but the game was very smooth on my rig which has 3 Geforce GTX 280's in 3-Way SLI. With PhysX enabled it ran well.
 
WabeWalker said:
I turned V-Sync off and ran the benchmark again. (This is the GTX 275/9800gt setup, with the 9800gt obviously handling PhysX).

The numbers were a bit different.

Min: 44
Avg: 75
High: 110

That's weird. My minimum was actually a bit higher with V-sync switched off? Also, my fans went ballistic. That didn't happen when V-sync was enabled.

Also, please note that I'm not saying that PhysX is the second coming or anything. I realize that very few games feature it prominently. I just thought that this might be worth noting.

One thing I can say is that I actually did notice a big difference at the beginning of the game when the 9800gt was handling the PhysX. This is the section of the game where The Joker is in the chair, and everyone's in that huge room, with all the flags flying above.

I could actually tell that the game was running more smoothly.

Before, when the GTX 275 was handling the PhysX by itself, that section of the game was just a little bit... I don't want to say stuttery... but somehow not quite smooth. When I returned to that part of the game, with the 9800gt enabled for PhysX, the game ran like butter. I mean, I could actually feel the difference. It was very noticeable.

Anyway, for whatever it's worth. There it is.
Click to expand...

Thanks for posting the numbers. Very nice performance increase with the dedicated PhysX card.

I have a spare 8800GT, you've convinced me to install it and use it as PhysX GPU, I hope my PSU can handle it!

Edit

What driver iteration are you using, 190.62?
 
Last edited:
meatfestival said:
Methinks I will hang on to my 9800GTX when I upgrade.
Click to expand...
by upgrade I hope you mean an entire upgrade as just a gpu upgrade would be a complete waste on that cpu. you arent even pushing that highly overclocked 9800gtx to more than 70-80% of what it can do in many games. ;)
 
cannondale06 said:
by upgrade I hope you mean an entire upgrade as just a gpu upgrade would be a complete waste on that cpu. you arent even pushing that highly overclocked 9800gtx to more than 70-80% of what it can do in many games. ;)
Click to expand...

Oh yes. Will be going for a Core i7.
 
WabeWalker said:
I have a single GTX 275 (along with a core i7), and I was shocked at the benchmark result. This was with everything set to the highest possible setting, including PhysX, and with 4X anti-aliasing, at 1920x1080.

I had been expecting better results than that.
Click to expand...

Look like decent results to me for a 275 with all those max settings and physics. If you had a 285 or 295 then maybe you could expect 100 fps, but we 250/260/275 people need to be more realistic.
 
cannondale06 said:
well I would hope so.:p
Click to expand...

Well it runs about the same whether I have PhysX enabled or not. I meant to add that to the post but since it was 4AM I was on the virge of passing out and i guess I didn't add the rest of the thought to the post.
 
Last edited:
I've set-up my old 8800GT as a PhysX card and the performance increase is big.

Now I can run 1680 x 1050 8x AA and all other settings maxed, including High PhysX with the following score.

image1jm.jpg


The game feels much smoother and I can now use high PhysX, before I was using medium and it was a chug fest. Break out that old card it's worth it.
 
Last edited:
I need to get an Obsidian 800D. Once I have that then I can make use of my 8800GT for that purpose and combine it with the 3 Geforce GTX 280's in my system. Though I'm pushing it on the power envelope. My UPS can't out put enough power to cover my computer with monitor at this point. It may not cover the computer if I do that.
 
My setup is an Intel C2D 3.33GHz with an Evga GTX 260 Core 216 Superclocked Edition 896MB card and an Evga 8800GTS 640MB as physX.

Without a physX card:
Min 24
Avg 41
Max 57

With a physX card:
Min 28
Avg 57
Max 81
 
This is maybe the one game where having that extra card really does impact the performance. If the game wasn't so damned good then it wouldn't be worth the bother - but I'll be damned if the game actually is worth the hassle. The PhysX effects are just so damned cool.

This game is a masterpiece. I'm pretty much blown away by it.
 
Spicey, I also 'felt' the difference when my 9800gt was handling the PhysX. I didn't need the benchmark to tell me that my minimum had jumped from 20 to a whopping 37, or that my average frames jumped dramatically as well.

I could feel it as soon as the game started. (It was pretty exciting that it actually worked.)
 
Dan_D said:
I didn't check my actual frame rates but the game was very smooth on my rig which has 3 Geforce GTX 280's in 3-Way SLI. With PhysX enabled it ran well.
Click to expand...

that is the reason why its smooth :p
 
Results for E8400 @ 4.1GHz, GTX260 clocked/unclocked, 8800GT clocked/unclocked for PhysX.
These are to test the effect of adding an 8800GT and what clocking each card gives when paired together.
edit: all tests are at 1080p res

All game settings to highest, vsync off, PhysX set to highest.
(both cards are mild overclock versions so slightly higher default clocks)

GTX260 clocked (720/1475, 1208), no PhysX card, PhysX set to highest.
AA Min, Max, Average
0xAA 34, 76, 56
2xAA 30, 69, 51
4xAA 29, 68, 50
8xAA 29, 63, 47
8QAA 24, 60, 38
16xAA 27, 64, 47
16QAA ran very very slow in places and crashed at the end.

GTX260 stock (626/1350, 1053), 8800GT stock (625, 1566, 900)
AA Min, Max, Average
0xAA 41, 108, 73
4xAA 41, 92, 67
8xAA 38, 93, 64
8QAA 32, 82, 54
16xAA 36, 90, 63
16QAA 30, 75, 50

GTX260 clocked (720/1475, 1208), 8800GT stock (625, 1566, 900)
AA Min, Max, Average
0xAA 46, 109, 81
16xAA 41, 96, 70
16QAA 35, 87, 58

GTX260 clocked (720/1475, 1208), 8800GT clocked a bit (723, 1753, 916)
AA Min, Max, Average
0xAA 46, 111, 81
16xAA 43, 100, 71
16QAA 36, 86, 58


Clocking the main graphics card reaps the biggest benefit by far.
Clocking the PhysX card has almost no effect, so a 8800GT looks to be easily powerful enough for this benchmark at least.


Overall pretty good results from a clocked GTX260 without an extra PhysX card.
With the 8800GT for PhysX...

Comparing 0xAA results:
the lowest framerate jumps 35%
the average framerate jumps 45%

Comparing 16xAA results: (sadly couldnt get a full 16QAA result)
the lowest framerate jumps 51%
the average framerate jumps 49%

The extra 8800GT certainly works!
 
Last edited:
Nenu said:
Results for E8400 @ 4.1GHz, GTX260 clocked/unclocked, 8800GT clocked/unclocked for PhysX.
These are to test the effect of adding an 8800GT and what clocking each card gives when paired together.
edit: all tests are at 1080p res

All game settings to highest, vsync off, PhysX set to highest.
(both cards are mild overclock versions so slightly higher default clocks)

GTX260 clocked (720/1475, 1208), no PhysX card, PhysX set to highest.
AA Min, Max, Average
0xAA 34, 76, 56
2xAA 30, 69, 51
4xAA 29, 68, 50
8xAA 29, 63, 47
8QAA 24, 60, 38
16xAA 27, 64, 47
16QAA ran very very slow in places and crashed at the end.

GTX260 stock (626/1350, 1053), 8800GT stock (625, 1566, 900)
AA Min, Max, Average
0xAA 41, 108, 73
4xAA 41, 92, 67
8xAA 38, 93, 64
8QAA 32, 82, 54
16xAA 36, 90, 63
16QAA 30, 75, 50

GTX260 clocked (720/1475, 1208), 8800GT stock (625, 1566, 900)
AA Min, Max, Average
0xAA 46, 109, 81
16xAA 41, 96, 70
16QAA 35, 87, 58

GTX260 clocked (720/1475, 1208), 8800GT clocked a bit (723, 1753, 916)
AA Min, Max, Average
0xAA 46, 111, 81
16xAA 43, 100, 71
16QAA 36, 86, 58


Clocking the main graphics card reaps the biggest benefit by far.
Clocking the PhysX card has almost no effect, so a 8800GT looks to be easily powerful enough for this benchmark at least.


Overall pretty good results from a clocked GTX260 without an extra PhysX card.
With the 8800GT for PhysX...

Comparing 0xAA results:
the lowest framerate jumps 35%
the average framerate jumps 45%

Comparing 16xAA results: (sadly couldnt get a full 16QAA result)
the lowest framerate jumps 51%
the average framerate jumps 49%

The extra 8800GT certainly works!
Click to expand...

Those numbers would explain why NVIDIA is going from SIMD to MIMD...
 
Just FYI, do not trust the benchmark for what in-game performance will be like. I have two GTX 280s and in SLI, they show the minimum framerate at 43. In single card mode with one running physx, the minimum framerate is 36.

Yet during actual gameplay, single card mode is remarkably better. In SLI I was getting frequent slowdowns generally in large rooms or places with a lot of smoke. In single card mode, I get pretty much constant 60fps.This is with the latest drivers and sli profiles.
 
Atech said:
And people called AGEIA crazy for making a dedicated PPU, saying that a GPU would be more than enough to run both graphics and physics...go wonder.
Click to expand...

Actually nVidia has it set up so the cpu has trouble doing it, making the gpu the superior physx processor.
Posted via [H] Mobile Device
 
JCDenton said:
Just FYI,

Yet during actual gameplay, single card mode is remarkably better. In SLI I was getting frequent slowdowns generally in large rooms or places with a lot of smoke. In single card mode, I get pretty much constant 60fps.This is with the latest drivers and sli profiles.
Click to expand...


Thats pretty much what Ive noticed with my setup..The scarecrow levels do bog down a bit though.
 
Atech said:
And people called AGEIA crazy for making a dedicated PPU, saying that a GPU would be more than enough to run both graphics and physics...go wonder.
Click to expand...

What Nvidia is doing with enabling its widely available, high market share hardware to do physics is nothing like what Ageia was doing with its single purpose, highly specialized unknown brand and hardware.
 
You must log in or register to reply here.
Back
Top