Some Interesting Batman GPU Results

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 have a similar setup to yours (i7 and GTX 275, everything stock).

Here's my benchmark with highest possible settings, including PhysX, and with 4X anti-aliasing, at 1680x1050:

Minimum = 29
Maximum = 60
Average = 48

It rarely dips below 40 when I play.
 
Last edited:
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

You do you where both AMD and Intel are going right?
 
So it actually works. Dedicating one card to PhysX had a pretty substantial impact the on the performance.
That's all fine and dandy and sounds neat but what are you gaining? Spending $100 on a second GPU for some extra fog effects and drapes? No thanks. I've seen that done on the CPU for years.
 
That's all fine and dandy and sounds neat but what are you gaining? Spending $100 on a second GPU for some extra fog effects and drapes? No thanks. I've seen that done on the CPU for years.
Bingo. And you know what? It's been perfectly satisfactory for my gaming experience. I've never once stopped in a game and said "you know what? I bet there's a way to make these drapes move with more fluidity." :rolleyes:
 
That's all fine and dandy and sounds neat but what are you gaining? Spending $100 on a second GPU for some extra fog effects and drapes? No thanks. I've seen that done on the CPU for years.


No you havn't, you have seen scripted cloth and scripted fog, be honest.
 
The point is that nobody gives a shit.

This is how technology moves forward
In 5 years time, if games had the same realism we have today, I can see you complaining that technology hasnt moved forward, so dont tell me you dont give a shit.
 
This is how technology moves forward
In 5 years time, if games had the same realism we have today, I can see you complaining that technology hasnt moved forward, so dont tell me you dont give a shit.
That's quite a tangent you went on from one comment. Because no one cares about dynamic cloth vs. scripted cloth, we all want gaming to stagnate and never improve? Do you honestly think before you press the "Submit Reply" button?

The point is, all these PhysX effects are gimmicks at best. If having a PhysX-capable GPU allowed you to blow something up and raze half the level - that would be sweet. But it doesn't. It's the difference between a banner rippling in the wind and a banner rippling in the wind more fluidly (whoop-de-doo). Of course, there's underhanded marketing bullshit like in Batman: Arkham Asylum in which they just flat-out remove the effect altogether to hoodwink customers into thinking PhysX is something more than it isn't.

Edios: "Alright, we are finished with these PhysX banner effects, now we need to add in some steam effects."
NVIDIA: "You can't render volumetric smoke without PhysX either." *slides large bag of money across the table*
Edios: *wink*
NVIDIA: *wink*

If developers weren't paid off and everything was equal, would PhysX have better physics presentation than going through the CPU? Sure, by design GPUs do some physics effects much better. Would it be different enough for people to give a shit? Nope, that's my point. Unless you go around through every level and poke each banner five times to see if it ripples differently, you wouldn't notice a damn difference in normal game play. If I had a choice to run higher AA or PhysX I'd run higher AA. If I had a choice to run higher (or better) AF than PhysX, I'd run better AF. It's the last thing on the list because it's inconsequential. The wonderful thing about the world is that due to the laws of physics, objects and the environment around us behave pretty predictably. That's why it's easy to run a script and get almost the same effect while focusing computing power on things that are much more critical to graphics quality and game immersion.

EDIT: The other thing is a lot of the PhysX effects are so overdone that they aren't realistic. A good example is the dynamic papers in Batman: Arkham Asylum. That is NOT how paper would behave, and looks more ridiculous than the papers just lying there stagnant.
 
Last edited:
This is how technology moves forward
In 5 years time, if games had the same realism we have today, I can see you complaining that technology hasnt moved forward, so dont tell me you dont give a shit.

the point wasn't that the realism was not important, it was that the method used to obtain it was not important. if scripted response work well enough and don't choke the performance like physx does then why bother? I have only played the demo but what little physx brings here is not worth the performance hit. I would much rather have the smother game play or better graphics for that type of cost.
 
That's quite a tangent you went on from one comment. Because no one cares about dynamic cloth vs. scripted cloth, we all want gaming to stagnate and never improve? Do you honestly think before you press the "Submit Reply" button?

The point is, all these PhysX effects are gimmicks at best. If having a PhysX-capable GPU allowed you to blow something up and raze half the level - that would be sweet. But it doesn't. It's the difference between a banner rippling in the wind and a banner rippling in the wind more fluidly (whoop-de-doo). Of course, there's underhanded marketing bullshit like in Batman: Arkham Asylum in which they just flat-out remove the effect altogether to hoodwink customers into thinking PhysX is something more than it isn't.

Edios: "Alright, we are finished with these PhysX banner effects, now we need to add in some steam effects."
NVIDIA: "You can't render volumetric smoke without PhysX either." *slides large bag of money across the table*
Edios: *wink*
NVIDIA: *wink*

If developers weren't paid off and everything was equal, would PhysX have better physics presentation than going through the CPU? Sure, by design GPUs do some physics effects much better. Would it be different enough for people to give a shit? Nope, that's my point. Unless you go around through every level and poke each banner five times to see if it ripples differently, you wouldn't notice a damn difference in normal game play. If I had a choice to run higher AA or PhysX I'd run higher AA. If I had a choice to run higher (or better) AF than PhysX, I'd run better AF. It's the last thing on the list because it's inconsequential. The wonderful thing about the world is that due to the laws of physics, objects and the environment around us behave pretty predictably. That's why it's easy to run a script and get almost the same effect while focusing computing power on things that are much more critical to graphics quality and game immersion.

EDIT: The other thing is a lot of the PhysX effects are so overdone that they aren't realistic. A good example is the dynamic papers in Batman: Arkham Asylum. That is NOT how paper would behave, and looks more ridiculous than the papers just lying there stagnant.

I'm not saying that there isnt some underhand stuff going on but I think you complain too much.
I think Batman AA looks fantastic, the complaints that the effects are 'unnecessary' are ridiculous.

If you would rather run higher AF, why not do it instead of moaning about how you cant, when you can?
I'm using an E8400 with GTX260 and found just that combo to give good performance in Batman AA at 1080p res with all quality options and PhysX maxed at decent AA levels. Adding an old 8800GT I have lying around improves it even further.
So it by no means needs killer hardware to get the effects.

Getting Physics right has to start somewhere yet it looks like you want to knock the advances, which.... "have" to start somewhere.
If you dont like the effects then dont use them.
Instead you create nuisance posts about having to use them when you dont.
Its very easy to not have a problem with PhysX effects.
"you doth complain too much".
 
the point wasn't that the realism was not important, it was that the method used to obtain it was not important. if scripted response work well enough and don't choke the performance like physx does then why bother? I have only played the demo but what little physx brings here is not worth the performance hit. I would much rather have the smother game play or better graphics for that type of cost.

I agree that things can be made to look nice using CPU Physics but it restrains developers from developing truly dynamic worlds and thus stunts the progress of realism in games.
Physics development needs to happen and is starting to look pretty good!
Some of the methods used arent very palletable, but thats another issue.

Note: I'm not having any problems getting Batman AA to run well on an E8400 + GTX260, no where near cutting edge and not expensive.
 
That's all fine and dandy and sounds neat but what are you gaining? Spending $100 on a second GPU for some extra fog effects and drapes? No thanks. I've seen that done on the CPU for years.

Only worth it if you have a decent spare GPU knocking about. ie you have just upgraded and haven't sold the spare card yet. Otherwise I wouldn't bother spending the money even though the performance increase is much improved
 
I agree that things can be made to look nice using CPU Physics but it restrains developers from developing truly dynamic worlds and thus stunts the progress of realism in games.
Physics development needs to happen and is starting to look pretty good!
Some of the methods used arent very palletable, but thats another issue.

Note: I'm not having any problems getting Batman AA to run well on an E8400 + GTX260, no where near cutting edge and not expensive.

what is the exact definition of "truly dynamic worlds"

or this kind of dynamic

http://vodpod.com/watch/1536922-velocity-physics-engine
 
I'm not saying that there isnt some underhand stuff going on but I think you complain too much.
I think Batman AA looks fantastic, the complaints that the effects are 'unnecessary' are ridiculous.

If you would rather run higher AF, why not do it instead of moaning about how you cant, when you can?
I'm using an E8400 with GTX260 and found just that combo to give good performance in Batman AA at 1080p res with all quality options and PhysX maxed at decent AA levels. Adding an old 8800GT I have lying around improves it even further.
So it by no means needs killer hardware to get the effects.

Getting Physics right has to start somewhere yet it looks like you want to knock the advances, which.... "have" to start somewhere.
If you dont like the effects then dont use them.
Instead you create nuisance posts about having to use them when you dont.
Its very easy to not have a problem with PhysX effects.
"you doth complain too much".
So because I disagree with you, I'm complaining? Are you five years old? I'm stating my opinion on the matter, quite like tens of other people in this thread have. If you can't deal with other opinions, unplug the ethernet cord.

PhysX ISN'T getting physics right. Marketing a proprietary solution by buying off devs to gimp their applications on other hardware is just this side of illegal. If you so cared about "the advancement of gaming," you'd condemn such actions and support OpenCL and the like. Furthermore, NVIDIA isn't actually DOING anything with PhysX - everything they add to games is a superficial gimmick at best. Why? Because no dev is stupid enough to base their game design (and really cool physics, if PhysX can indeed deliver) on a proprietary solution.
 
The trouble with the PhysX API, is that, even though it offloads work to the GPU, it still sucks. Sure there are gimmicky things like cloth and particle effects, but there is nothing that makes you say "wow!' or meaningfully contributes to gameplay no matter how hard they try. The actual implementation of the effects is also lacking. Havok, while being CPU powered, has much smoother, far more convincing collisions than PhysX, and PhysX's deformation and cloth physics are just ugly. The accuracy of physics implementation is much more important than the extent of physics implementation. The fact that PhysX is so wanting in accuracy, while using such a vastly huger amount of processing power than Havok, suggests to me than it should be abandoned post haste.

I do look forward a time when physics plays a bigger role in our games, but I don't thing that halving your frame rate for a few buggy flags and inexplicably large numbers of apparently hollow styrofoam crates is what we're looking for. I don't think developers should even consider PhysX until it can do a good a job as Havok did in Half Life 2 at acceptable performance levels.

This is not even taking in consideration that fact that it is proprietary, which is a huge issue unto itself.


No you havn't, you have seen scripted cloth and scripted fog, be honest.

Yeah, but the unscripted cloth and fog looks so bad that the scripted version is preferable.
 
So because I disagree with you, I'm complaining? Are you five years old? I'm stating my opinion on the matter, quite like tens of other people in this thread have. If you can't deal with other opinions, unplug the ethernet cord.

PhysX ISN'T getting physics right. Marketing a proprietary solution by buying off devs to gimp their applications on other hardware is just this side of illegal. If you so cared about "the advancement of gaming," you'd condemn such actions and support OpenCL and the like. Furthermore, NVIDIA isn't actually DOING anything with PhysX - everything they add to games is a superficial gimmick at best. Why? Because no dev is stupid enough to base their game design (and really cool physics, if PhysX can indeed deliver) on a proprietary solution.

Dude, you are entitled to your opinion but when you make it you shouldnt mix up different issues.
PhysX is an evolutionary step as already mentioned and thank f someone is moving physics simulations forward at pace.
I didnt say it was done right, was the first rocket ever made done right, or the one ofter that.... ?
if you want perfection from the start, an even larger jump in hardware requirements will be needed and years more of development.
Surely then it needs to start making some progress?
Good job it is :)

If NVidia are doing something illegal it will come home to roost at some point.
To me this is just standard marketing and is no surprise, take for example Crysis and DX10.
It was later found that practically all of the "DX10" effects can be enabled in DX9.
In a very similar fashion, all the PhysX effects can be enabled to run on CPU.
Unfortunately it looks like it requires a strong CPU whereas hardware PhysX doesnt, so there is value in the NVidia solution still.
 
The trouble with the PhysX API, is that, even though it offloads work to the GPU, it still sucks. Sure there are gimmicky things like cloth and particle effects, but there is nothing that makes you say "wow!' or meaningfully contributes to gameplay no matter how hard they try. The actual implementation of the effects is also lacking. Havok, while being CPU powered, has much smoother, far more convincing collisions than PhysX, and PhysX's deformation and cloth physics are just ugly. The accuracy of physics implementation is much more important than the extent of physics implementation. The fact that PhysX is so wanting in accuracy, while using such a vastly huger amount of processing power than Havok, suggests to me than it should be abandoned post haste.

I do look forward a time when physics plays a bigger role in our games, but I don't thing that halving your frame rate for a few buggy flags and inexplicably large numbers of apparently hollow styrofoam crates is what we're looking for. I don't think developers should even consider PhysX until it can do a good a job as Havok did in Half Life 2 at acceptable performance levels.

This is not even taking in consideration that fact that it is proprietary, which is a huge issue unto itself.




Yeah, but the unscripted cloth and fog looks so bad that the scripted version is preferable.

http://www.youtube.com/watch?v=_ZfXIFFmB7M

look at 7:30... is that really water effect or is it potato rolling on the table :p

sorry for physX lover, the thing just not working well... :rolleyes:
 
http://www.youtube.com/watch?v=_ZfXIFFmB7M

look at 7:30... is that really water effect or is it potato rolling on the table :p

sorry for physX lover, the thing just not working well... :rolleyes:

Wow, honestly I haven't played Cryostasis yet, because I heard it was so shitty, but that water does look kinda crappy on that table (and I have Nvidia cards).

I really can't say anything (bad) about Batman:AA, as I get constant 60fps everywhere. It's worth the 50 bucks. I really can't say anything (good or bad) about the PhysX, all we do is sit and watch his cape all day.
 
Last edited:
The trouble with the PhysX API, is that, even though it offloads work to the GPU, it still sucks. Sure there are gimmicky things like cloth and particle effects, but there is nothing that makes you say "wow!' or meaningfully contributes to gameplay no matter how hard they try. The actual implementation of the effects is also lacking. Havok, while being CPU powered, has much smoother, far more convincing collisions than PhysX, and PhysX's deformation and cloth physics are just ugly. The accuracy of physics implementation is much more important than the extent of physics implementation. The fact that PhysX is so wanting in accuracy, while using such a vastly huger amount of processing power than Havok, suggests to me than it should be abandoned post haste.

I do look forward a time when physics plays a bigger role in our games, but I don't thing that halving your frame rate for a few buggy flags and inexplicably large numbers of apparently hollow styrofoam crates is what we're looking for. I don't think developers should even consider PhysX until it can do a good a job as Havok did in Half Life 2 at acceptable performance levels.

This is not even taking in consideration that fact that it is proprietary, which is a huge issue unto itself.

Yeah, but the unscripted cloth and fog looks so bad that the scripted version is preferable.
My sentiments exactly. Well said :cool:.

Dude, you are entitled to your opinion but when you make it you shouldnt mix up different issues.
PhysX is an evolutionary step as already mentioned and thank f someone is moving physics simulations forward at pace.
I didnt say it was done right, was the first rocket ever made done right, or the one ofter that.... ?
if you want perfection from the start, an even larger jump in hardware requirements will be needed and years more of development.
Surely then it needs to start making some progress?
Good job it is :)

If NVidia are doing something illegal it will come home to roost at some point.
To me this is just standard marketing and is no surprise, take for example Crysis and DX10.
It was later found that practically all of the "DX10" effects can be enabled in DX9.
In a very similar fashion, all the PhysX effects can be enabled to run on CPU.
Unfortunately it looks like it requires a strong CPU whereas hardware PhysX doesnt, so there is value in the NVidia solution still.
What issues do you think I'm mixing up? But hey, however low your standards are, or however you justify your being a fanboy, good for you.
 
I agree that things can be made to look nice using CPU Physics but it restrains developers from developing truly dynamic worlds and thus stunts the progress of realism in games.
Physics development needs to happen and is starting to look pretty good!
Some of the methods used arent very palletable, but thats another issue.

Note: I'm not having any problems getting Batman AA to run well on an E8400 + GTX260, no where near cutting edge and not expensive.

I hear you but I think you missed part of my point, until it actually does truly impact game play its just way too expensive in terms of the resources allocated, not so much in $ but in other effects and eye candy. for instance the OP almost doubled his avg frame rate by turning it off, and to me 35 with the low in the 20s is just no playable. and for what physx brought to the game its not even worth lowering the AA to use it. on the other hand I have all kinds of CPU power left over, it would not really affect my frame rates to use some of that.
 
I think we are a number of years away from PhysX or API's like it actually contributing to gameplay in a meaningful way. The technology is really in it's infancy compared to what it will look like in several years from now. Just because the technology doesn't contribute to game play in today's games doesn't mean that it should be dropped or done away with. It creates some cool effects and I think that's a good enough start for the time being at least. If game developers shy away from the technology because a bunch of geek's whine about not having the hardware to support it, or they bitch because it doesn't do enough yet then the technology might die off. I think in the long run, that would be most unfortunate. One day I think the technology will prove itself and provide ground breaking and game changing experiences but again we are a long way off from that day. If you can't see where this type of API could eventually lead then I think you are being just a bit short sighted.
 
I think we are a number of years away from PhysX or API's like it actually contributing to gameplay in a meaningful way. The technology is really in it's infancy compared to what it will look like in several years from now. Just because the technology doesn't contribute to game play in today's games doesn't mean that it should be dropped or done away with. It creates some cool effects and I think that's a good enough start for the time being at least.If game developers shy away from the technology because a bunch of geek's whine about not having the hardware to support it, or they bitch because it doesn't do enough yet then the technology might die off. I think in the long run, that would be most unfortunate. One day I think the technology will prove itself and provide ground breaking and game changing experiences but again we are a long way off from that day. If you can't see where this type of API could eventually lead then I think you are being just a bit short sighted.

hehe
QFT ;)

technology moves forward step by step, not by a huge friggin' leap
 
If game developers shy away from the technology because a bunch of geek's whine about not having the hardware to support it, or they bitch because it doesn't do enough yet then the technology might die off.

I think a lot of those "geeks that whine about not having the hardware to support it" either has the hardware or can afford to buy it if it matters enough. Most of the bitching is about it not doing enough (ingame physics instead of eyecandy), the removal of everything (banners in Mirrors edge was removed instead of having scripted or static) showing that games are not optimized due to being a marketing tool for GFX cards instead. Some might have prefered that they used Havok cloth and such that would run on all platforms.
What I think many of us geeks wants, is hardware accelerated ingame physics that are hardware agnostic and doesn't tell us what we should buy. Bullets OpenCL and open source might be a good candiate for that and so do Havok's opencl.
People would have accepted and liked PhysX better if it was to be ported to Opencl I think as hardware agnostic middleware. I don't mind Nvidia adding features to sell their cards (since I am a consumer, its always good). I do mind if they make developers do less for consumers in order to sell Nvidia hardware and not having it optimized for others on CPU:
http://www.hardforum.com/showpost.php?p=1034626406&postcount=8

Instead, I want scalable physics that work on all hardware, like this:
http://www.youtube.com/watch?v=cfSBHJRU9_Q
 
I think we are a number of years away from PhysX or API's like it actually contributing to gameplay in a meaningful way. The technology is really in it's infancy compared to what it will look like in several years from now. Just because the technology doesn't contribute to game play in today's games doesn't mean that it should be dropped or done away with. It creates some cool effects and I think that's a good enough start for the time being at least. If game developers shy away from the technology because a bunch of geek's whine about not having the hardware to support it, or they bitch because it doesn't do enough yet then the technology might die off. I think in the long run, that would be most unfortunate. One day I think the technology will prove itself and provide ground breaking and game changing experiences but again we are a long way off from that day. If you can't see where this type of API could eventually lead then I think you are being just a bit short sighted.

my sentiments exactly. i don't think anyone would say that hardware accelerated physics has nothing to offer; just like one might have seen the potential of 3d accelerator cards fifteen years ago. i think a good situation would be where a game can scale with cpu physics depending on how many cores a user has, and then offload additional processing onto the gpu if performance degrades and with heavier physics processing that the cpu can't handle at all - hence something like opencl physics i imagine (which has been mentioned earlier). the only issue with that is since the baseline is still the cpu, there still wouldn't be enough processing power (in relation to a gpu, comparatively speaking) to move beyond just more advanced physics effects to advanced gameplay physics (at least not at a rapid enough pace to make it particularly meaningful, imo). this would be even more apparent now since most are still on dualcore systems (not to mention consoles), which means unless a dev is willing to limit the potential sales to an absolute minimum, there probably won't be any ground breaking experiences as you suggest for some time to come.

as far as what we have now - and i mean right at this moment - is software based physics where the majority of games use either havok or physx. i think ageia's ppu and gpu physx were and are a necessary step forward. sure the situation hasn't worked out perfectly, but this isn't a perfect world. however, the awareness that the technology has brought forth into the marketplace is the progress that was necessary for new tools to become available that can cause game design & creativity to improve and potentially grow to new heights.

yes, the implementation can be considered lackluster thus far by some, but what can one expect from a dev if it means losing potential sales? unless one is willing to take the enormous risk of making something potentially amazing, but will only allow a few people to play it. from a business viewpoint, it would be financial suicide. or unless it was perhaps a small indie developed title that was built to truly showcase the technology - that might be the only feasible outcome if something were to happen. so even if/ when eventually hardware accelerated physics becomes available on all hardware, most games will still cater to the lowest common denominator, meaning consoles. perhaps meaning we may have to wait until the "next generation" arrives for that to start to take place. otherwise, perhaps a pc exclusive will be made to showcase advanced gpu accelerated gameplay physics eventually.

as far as the differences between physx and havok, i'm sure a game could be made to look exactly the same using either if it was practical. tech demos are one thing but actual games are a different story. sure in the games that are being complained about in regards to gpu physx, could there be better execution/ optimization? always. it's not simply just a case of saying, well this game looks and does so and so much better than another without regard to the methodology behind it and the time and effort that went into it. furthermore, we don't know if whatever gpu physx effects that were implemented in a game like mirrors edge or batman aa were ever intended to be included in the game in the first place. usually it's a case of the game just being designed on a multiplatform basis, using consoles as the "l.c.d." (lowest common denominator). then perhaps the devs may have had a bit extra time towards the end of development to include the gpu accelerated physx effects as an afterthought (and without even bothering to delay the game further to enable cpu fallback effects), which may partially explain the delays of the aforementioned games in comparison to the console versions. this may be similar to dirt2 being delayed on the pc for dx11 support. imagine what could be done if more time and painstaking effort were taken to make them more meaningful to the game as a whole.

so i don't think it's a matter of, well this could have been done on the cpu with less particles or static effects or scripted animations, but more of the fact that perhaps the intention was never to have included them at all from the get-go when development was already in full swing with all assets in place. as a result, one can still wish to speculate that nvidia threw money to push a dev to integrate gpu physx effects in a game - which is fine - but that doesn't mean that the dev would have bothered in the first place to integrate said effects at all, whether via software or otherwise.

consider older games like graw1/2 and the accelerated physics effects. the games were designed to run on a lowly p4 cpu in terms of system requirements (oddly enough, the system requirements for batman aa only requires a single core cpu to run as well). so it may have been a similar situation where the game was designed for the "l.c.d." and the accelerated effects were added later on in development. could some of the effects have been made scaleable? of course! but one can't deny this is a relatively new arena for game developers, and it will certainly take some time for the technology to mature even further.

we will most certainly see how things turn out in regards to physics acceleration over the next couple of years and how the companies involved in its development and the competitive forces coming into play continue to shape the future of gaming. until then, let the fanboy wars ensue.
 
Last edited:
Back
Top