How much of a role will physx play

[Wreckage]

I played games on Dos too,Win95 set the way for easier gaming and start of DX standard.

I don't remember DX getting good until version 9.

Besides there was already OpenGL.

 

[Mem]

I don't remember DX getting good until version 9.

Besides there was already OpenGL.

You are missing the point,start of DX standard back in Win95 days started the building blocks for DX and Windows gaming,plus making it easier for gamers.

 

[nobodyknows]

Owning a card without PhyX is like owning a card without DirectX.

Now here is a Nvidia fan.

With multi-core cpu's becoming the standard it seems to me to make more sense for the gaming industry to take advantage of that rahter then to favor one GPU manufacturer over the other?

In the long run competition is good for the consumer.

 

[Creig]

Owning a card without PhyX is like owning a card without DirectX.

It's ironic that you would choose to compare PhysX to DirectX, given their completely opposite adoption strategies:

The DirectX programming team worked very hard to ensure it would run on as many different hardware configurations as possible.

That is why DirectX succeeded.

Nvidia, on the other hand, has decided to artificially limit PhysX usage exclusively to its own hardware, even though it has been proven to work just fine when using an ATi video card as the primary display.

That is why PhysX is failing.

 

[Dribble]

DirectX is the PC standard these days but think back...

DirectX had glide and opengl as competion. Glide was locked to one graphics card manufacturer. DirectX was locked to one OS (win95). OpenGL worked on everything - it is the perfect example of an open standard. Not only that it *was* the standard across the board - everything ran openGL.

Hence trying to argue how directx is so much more open and better then physx is stupid. It has mostly killed the very good open standard. It's also a good example of open standards not winning.

You may disagree with Wreckage because he is wreckage, but he has succeeded in trying to get you to defend directx, a closed standard, hence contradicting the whole basis of your argument.

Ending with the line "That is why PhysX is failing" is silly too - the problem is that it isn't. It is the most popular software standard, and by default the most popular gpu accelerated standard (there is only one).

I too think Wreckage is an argumentative fan boy, but like Charlie there is elements of truth in what he says.

 

[Wreckage]

That is why DirectX succeeded.

DirectX succeeded because of the windows monopoly.

 

[AzN]

I don't remember DX getting good until version 9.

Besides there was already OpenGL.

There was actually glide. OPENGL was being touted by carmack within their circle but it never really took off.

 

[Daedalus685]

DirectX succeeded because of the windows monopoly.

Thus physX won't succeed because of the lack of said monopoly? :awe:

Windows had nothing to do with ATI, nvidia, and devs opting not to go with exclusively open GL cards..

 

[Daedalus685]

There was actually glide. OPENGL was being touted by carmack within their circle but it never really took off.

heh, if not for carmack open GL would be about as known as all those pokemon rip offs from a decade ago

 

[Creig]

DirectX is the PC standard these days but think back...

DirectX had glide and opengl as competion. Glide was locked to one graphics card manufacturer. DirectX was locked to one OS (win95). OpenGL worked on everything - it is the perfect example of an open standard. Not only that it *was* the standard across the board - everything ran openGL.

Hence trying to argue how directx is so much more open and better then physx is stupid. It has mostly killed the very good open standard. It's also a good example of open standards not winning.

You may disagree with Wreckage because he is wreckage, but he has succeeded in trying to get you to defend directx, a closed standard, hence contradicting the whole basis of your argument.

I'm going to go out on a limb here and assume you are a major proponent of either Linux or Apple. :hmm:

Given its overwhelming worldwide popularity, Windows is a standard unto itself and my comments were phrased taking that assumption into account. I don't usually need to specify the OS I'm referring to when posting comments in the Video Forum.

So within the Windows platform, DirectX has attempted to become hardware agnostic, running on as many hardware vendors products as possible whereas Nvidia has limited PhysX to only those systems utilizing an Nvidia video card for the primary display.

Ending with the line "That is why PhysX is failing" is silly too - the problem is that it isn't. It is the most popular software standard, and by default the most popular gpu accelerated standard (there is only one).

I too think Wreckage is an argumentative fan boy, but like Charlie there is elements of truth in what he says.

PhysX is failing in the respect that it could have easily become the defacto GPU accelerated physics processing software on all Windows computers if they had opened it up to as many systems as possible, ala DirectX. But instead, they have deliberately and drastically limited the number of computers it can function on, thus adoption has been extremely low by both developers and end users alike. This is also the reason many people, like myself, hope that a completely hardware agnostic solution will emerge and quickly supplant PhysX.

Until a platform comes out that will run well on both ATi AND Nvdia cards, GPU accelerated physics processing growth will remain stunted

 

[NoQuarter]

I don't remember DX getting good until version 9.

Besides there was already OpenGL.

DirectX was decent at 6, it was good at 7. The reason DirectX was adopted was as Creig said, the developers worked hard to ensure hardware compatibility and outpace OpenGL on features while working closely with hardware vendors and software developers to add features they wanted.

As for DirectX being a closed standard, I'm only interested in what will advance the gaming technology the most. Nvidia controlling PhysX is nice for short term progress but terrible for long term progress.

If Ageia still owned PhysX at least their PPU could have failed and they could then write GPU accelerated PhysX extensions in both CUDA and Stream, then OpenCL and DirectCompute, and licensed all this crap to everyone. Remember PhysX used to have the highest licensing fees when Ageia owned it that's where they made all their money.. they would have been all over this shit.

But since Nvidia owns it we get a few added Nvidia effects and the platform itself stagnates because noone can play ball with them.

 

[Wreckage]

PhysX is failing in the respect that it could have easily become the defacto GPU accelerated physics processing software on all Windows computers if they had opened it up to as many systems as possible, ala DirectX.

Uh they currently are the defacto GPU accelerated physics processing software.

They did offer it to AMD. So NVIDIA is not the problem here.

Not to mention that software PhysX runs on more than just Windows unlike DirectX. So it's more of an open cross platform standard. :thumbsup:

 

[Creig]

Uh they currently are the defacto GPU accelerated physics processing software.

Sorry, I should have said "widely adopted", which PhysX most definitely is not.

They did offer it to AMD. So NVIDIA is not the problem here.

So what is keeping Nvidia from removing the block they placed within PhysX to prevent it from working in any system with an ATi video card? I didn't realize they needed AMD's permission to do so.

Nvidia is the problem here.

Not to mention that software PhysX runs on more than just Windows unlike DirectX. So it's more of an open cross platform standard. :thumbsup:

As you could see from my comments, I was addressing hardware GPU based physics processing on Windows computers, not software physics nor consoles. Those are completely different cans of worms.

 

[MagickMan]

Uh they currently are the defacto GPU accelerated physics processing software.

And PhysX sucks compared to CPU-based physics, so that isn't saying much. It's like saying you're the biggest ball of shit in the toilet.

 

[Fox5]

DirectX is the PC standard these days but think back...

DirectX had glide and opengl as competion. Glide was locked to one graphics card manufacturer. DirectX was locked to one OS (win95). OpenGL worked on everything - it is the perfect example of an open standard. Not only that it *was* the standard across the board - everything ran openGL.

Hence trying to argue how directx is so much more open and better then physx is stupid. It has mostly killed the very good open standard. It's also a good example of open standards not winning.

You may disagree with Wreckage because he is wreckage, but he has succeeded in trying to get you to defend directx, a closed standard, hence contradicting the whole basis of your argument.

Ending with the line "That is why PhysX is failing" is silly too - the problem is that it isn't. It is the most popular software standard, and by default the most popular gpu accelerated standard (there is only one).

I too think Wreckage is an argumentative fan boy, but like Charlie there is elements of truth in what he says.

OpenGL was not supported on all hardware back in the day. ATI has only had good opengl support in recent times, and 3dfx never had it. Nvidia did, but for a while with all the proprietary extensions nvidia was practically pushing opengl as its own api.

Not to mention how bad 3d drivers were for Linux, and Mac was never much of a gaming platform. In regards to the gaming market, it didn't matter (at the time) that OpenGL was supported in more operating systems. Linux didn't really take shape as a decent desktop OS until post year 2000, and Mac was on the verge of death until OSX.

Still, the short period of time where opengl was a superior api to directx (and glide was out of the picture), it only worked well on nvidia cards.

 

[Dark Shroud]

I'm also pretty sure that Havok is used in more PC games than PhysX.

And with Intel as their sugar daddy as well as actually working with companies they appear to have a longer life span than PhysX.

 

[v8envy]

Owning a card without PhyX is like owning a card without DirectX.

I own a cards *with* PhysX. Do you know how often I've enabled PhysX in the last decade?

Approximately zero times.

On the other hand, I have used the DirectX capabilities every day on cards with or without PhysX support. So, in my case and in the case of anyone actually playing PC games the two are of vastly different levels of importance.

So yeah, once again, you're utterly full of crap.

 

[Wag]

I'm also pretty sure that Havok is used in more PC games than PhysX.

And with Intel as their sugar daddy as well as actually working with companies they appear to have a longer life span than PhysX.

http://physxinfo.com/

A lot more games use it than you think (even though they might not be the GPU kind).

 

[Seero]

I guess people are confused about PhysX not running on platforms that doesn't have Nvidia hardware in it. PhysX simply runs off CPU. A Nvidia card have a capability to accelerate (offloading from CPU) PhysX codes. Nvidia didn't disable PhysX upon ATI video card, but disable PhysX the on the Nvidia card. That means, you can't use an ATI card for graphics and a Nvidia card for PhysX. Shall you have such setup, programs that contains PhysX code will simply run on CPU.

People are claiming that Havok is better than PhysX without knowing that PhysX also run on CPU. Now does PhysX run as good as Havok when it is running on CPU only? So far no one tested it. Havok did had a plan to offload physics computation onto GPU (called Havok FX), but sadly abandoned.

 

[NoQuarter]

http://physxinfo.com/

A lot more games use than you think (even though they might not be the GPU kind).

Yea I counted the listed games on the Physx site and Havok site and it was something like 220 vs 240, pretty close both are really popular. They are both fine engines, the GPU accelerated application of PhysX is just spotty.

 

[zerocool84]

Uh they currently are the defacto GPU accelerated physics processing software.

They did offer it to AMD. So NVIDIA is not the problem here.

Not to mention that software PhysX runs on more than just Windows unlike DirectX. So it's more of an open cross platform standard. :thumbsup:

Yea Physx is soooooooooo open. How many ATI cards can it run on??? Can you have an ATI card and an Nvidia card as a Physx card??? Yea sounds sooooooooooooo open to me.

 

[Wag]

http://international.download.nvidia.com/downloads/cool_stuff/demos/SupersonicSledInstaller1.0.exe
http://international.download.nvidia.com/downloads/cool_stuff/demos/DesignGarageInstaller1.0.msi

BTW: If anyone cares Nvidia just released the SuperSonic Sled demo (won't run unless you have a 4xx card) and Design Garage Demo (which is supposed to run on 2xx or better but I got an error on when I tried to run on my 295).

 

[Dribble]

It's ironic that you would choose to compare PhysX to DirectX, given their completely opposite adoption strategies:

The DirectX programming team worked very hard to ensure it would run on as many different hardware configurations as possible.

That is why DirectX succeeded.

Nvidia, on the other hand, has decided to artificially limit PhysX usage exclusively to its own hardware, even though it has been proven to work just fine when using an ATi video card as the primary display.

That is why PhysX is failing.

Actaully phsyx runs on pretty well every hardware platform - all the consoles, pc's, macs (I think), even iphones. It's just only gpu accelerated on nvidia cards on the pc.

That gpu accelerated physx I would say actually lives or dies on console support not graphics card support. It's such a successful software package because nvidia basically handed it out for free to all the consoles, and actively support it. If the consoles support it then it ends up in all the console ports, which these days is most PC games.

Consoles don't support gpu accelerated physics so hence all the console ports don't either. Even if nvidia developed it for ati cards for free it wouldn't really get adopted, same as DX10 and 11 haven't really been adopted (I know games with DX10 and 11 stickers exist but nearly all of them look near identical in DX9 - they're really DX9 games with a couple of minor DX10/11 additions).

Hence whether phsyx fails or succeeds is basically dependent on what happens in the next round of consoles. Assuming they have the hardware for hardware physics support (very very likely) then there will be competition to become the accepted standard. This I assume is also what hardware accelerated physics libraries like bullet are really aiming at. Whether physx wins or looses that is dependent on what nvidia do - I wouldn't be surprised if it miraculously gets ported to open cl for PS4 and directX compute for xbox 720 at that time.

 

[NoQuarter]

Actaully phsyx runs on pretty well every hardware platform - all the consoles, pc's, macs (I think), even iphones. It's just only gpu accelerated on nvidia cards on the pc.

That gpu accelerated physx I would say actually lives or dies on console support not graphics card support. It's such a successful software package because nvidia basically handed it out for free to all the consoles, and actively support it. If the consoles support it then it ends up in all the console ports, which these days is most PC games.

Consoles don't support gpu accelerated physics so hence all the console ports don't either. Even if nvidia developed it for ati cards for free it wouldn't really get adopted, same as DX10 and 11 haven't really been adopted (I know games with DX10 and 11 stickers exist but nearly all of them look near identical in DX9 - they're really DX9 games with a couple of minor DX10/11 additions).

Hence whether phsyx fails or succeeds is basically dependent on what happens in the next round of consoles. Assuming they have the hardware for hardware physics support (very very likely) then there will be competition to become the accepted standard. This I assume is also what hardware accelerated physics libraries like bullet are really aiming at. Whether physx wins or looses that is dependent on what nvidia do - I wouldn't be surprised if it miraculously gets ported to open cl for PS4 and directX compute for xbox 720 at that time.

I think everyone in the thread is referring to PhysX as shorthand for GPU-accelerated PhysX but full well knows the difference (I'd hope).

I definitely agree that consoles are going to dictate the progression of GPU-accelerated physics at this point, but MS and Nintendo have already commited to ATI again and Sony is looking for a partner other than Nvidia for their next console.

This leaves no CUDA enabled GPU's in the consoles but I would imagine MS will push DirectCompute in their next console, and I'm sure Nintendo and to some degree MS will also want to use OpenCL.

So I see them having to port GPGPU PhysX extensions to another language come the next console wave.

 

[Silverforce11]

Games have a long development time, so there will be some released this year with Physx. However, the next batch of games is unlikely to given that directx11 can do physics with direct compute, or use havok which scales well on CPUs.

The only game where physx really made a difference was mirrors edge. All the others are just fluff, and on mid range cards (or even gtx275 ), will tank performance down hard with physx on and max image quality settings.

Metro is the only big title recently to have physx, and its pretty much irrelevant because with physx OFF, even a GTX 480 suffers horrible FPS with high settings and no AA. With it on, you can enjoy your slideshow.

Its on the out. Devs won't bother waste precious development time to add fluff that doesn't work for half of their customers, unless NV pays them $$. Even then, the final result is "meh" in most games.

Heck, look at BF:BC2, so much destructibles and pretty eye candy, i'm surprised NV didnt manage to convince DICE to add in physx somewhere. Maybe not enough $$ to throw around these days.