NVIDIA, Epic add DX11 features to Unreal Engine 3

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

itsmydamnation

Diamond Member
Feb 6, 2011
3,045
3,835
136
Not this again...do I need to dig up the thread about tesselation again?! :eek:

Thesselation is already "dynamic".
On the fly.

The only resaon for that setting in AMD' driver is due to lack of tesselation performance.

This is just the same old broken tune from AMD fans in order to jusitfy the lower performance. :thumbsdown:

no its about balance, so long as rasterization works on pixel quads, and a maximum of one triangle per quad then the higher the level of tessalation the lower the performance you get on rasterization. that means you need more ROP's that consumer more power for extremely limited benifit.

to me Tessalation offers the best advanage going in the other direction, use it as a replacement for LOD. far objects render at 1X and as you get closer is scales up, will be far more seemless then crappy lod we have to put up with now.


Think about it.
And the apply it to other features.

Too much AF...
Too much AA...
To much...performance.

Hillarious.

there is a big difference, these things dont directly impact the performance of other parts of the GPU, they might effect bandwdith utilization but they dont directly change other parts of the GPU's workload while Tessalation has a very big impact on that.
 
Last edited:

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
Asking "Why should Physx be run on the GPU?" is like asking "Why should games be hardware accelerated?" At least before DirectX became dominant, games could be run on the CPU. They just ran a whole lot faster on hardware.

Now don't you dare use logic in a debate...shame on you.

I have asked people the same thing (when it was both NVIDIA and AMD fans trying to downplay PhysX, back when AGEIA owned PhysX) if they thought their multicore CPU could hold a candle to their GPU(or a PPU) when it came to rendering graphics.

Today the NVIDIA side has seen the light, but the AMD side still refuses.

Which to me shows that the resistance is only driven by brand loyality...not facts.

Just wait and see...when AMD get's (if ethey ever do that is) their own GPU implemenation, the "CPU argument" will die and people will pretend it never really mattered...just wait and see :D
 

Skurge

Diamond Member
Aug 17, 2009
5,195
1
71
It mattered to AMD pre-FERMI:
http://blogs.amd.com/play/2009/06/02/why-we-should-get-excited-about-directx-11/

The sky was the limit.

After FERMI, PR spin control went into action:
http://blogs.amd.com/play/2010/11/29/tessellation-for-all/

It'e really thit simple:

When AMD thought they would have the egde due to their previous dabbles in tesselation it was the most important factor of DX11.

When AMD realized their performance was subpar to NVIDIA's there suddenly was a thing as "too much tesselation"...

Think about it.
And the apply it to other features.

Too much AF...
Too much AA...
To much...performance.

Hillarious.

And yet AMD keeps helping Devs Add tessellation to games. :hmm:

Yes, you go ahead and listen to PR. I'll look at real world benchmarks. Even unigene that isn't even a game with normal tessellation which is more than any game out there and AMDs cards are just as fast as nvidia's. So it doesn't matter to me who has more tessellation power if its not going to be used.

So yeah, you keep listening to all the PR both companies spew out. I think I remember nVidia saying DX11 didn't matter when they had no DX11 hardware.

Yeah, go ahead and run 64xAA or 128xAF and see what happens.
 
Last edited:

Skurge

Diamond Member
Aug 17, 2009
5,195
1
71
Now don't you dare use logic in a debate...shame on you.

I have asked people the same thing (when it was both NVIDIA and AMD fans trying to downplay PhysX, back when AGEIA owned PhysX) if they thought their multicore CPU could hold a candle to their GPU(or a PPU) when it came to rendering graphics.

Today the NVIDIA side has seen the light, but the AMD side still refuses.

Which to me shows that the resistance is only driven by brand loyality...not facts.

Just wait and see...when AMD get's (if ethey ever do that is) their own GPU implemenation, the "CPU argument" will die and people will pretend it never really mattered...just wait and see :D

Again, I don't care where the physx are run, I even bought an nVidia card to try it out FFS. Its poorly done, it brings nothing to the table. The numbers behind it must be huge, but it doesnt translate to anything useful while playing the game. I don't have a problem with GPU physics, I have a problem with the GPU accelerated physx's poor execution. It does't look or perform and make the game anymore fun than CPU accelerated physx, but I need more hardware to use it? Screw that.
 

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
no its about balance, so long as rasterization works on pixel quads, and a maximum of one triangle per quad then the higher the level of tessalation the lower the performance you get on rasterization. that means you need more ROP's that consumer more power for extremely limited benifit.

to me Tessalation offers the best advanage going in the other direction, use it as a replacement for LOD. far objects render at 1X and as you get closer is scales up, will be far more seemless then crappy lod we have to put up with now.

Lets see what AMD has to say:

1. A beast called the tessellator has been added which enables games developers to create smoother, less blocky and more organic looking objects in games. This is the change you’ll probably be most aware of. And it’ll show up when you look at the silhouettes of hills and mountains or the profiles of characters in games. Where artists previously had to trade off quality for performance, now artists will have the freedom to create naturalistic scenery. We’ve gotten used to seeing strangely blocky ears and noses on our opponents. But the new generation of games should allow those opponents to scare the heck out of us instead. The tessellator represents a natural next step in gaming hardware (in fact the Xbox 360 graphics chip that AMD designed already has a tessellator, and AMD graphics hardware has featured tessellator technology starting with the ATI RadeonTM HD 2000 series right up to the latest ATI RadeonTM HD 4000 series cards today).

That was pre-FERMI.

Before they got beaten in tesselation.

What you are suggestion is just the defeated trying to justify the performance deficit of their fixed function tesselation engine.

Like I said:
Hillarious.


there is a big difference, these things dont directly impact the performance of other parts of the GPU, they might effect bandwdith utilization but they dont directly change other parts of the GPU's workload while Tessalation has a very big impact on that.

And?
If you got the performance (NVIDIA's unified solution) it still beats the crap out of AMD's fixed function solutions

So you are advocating the slower solution...talk about holding technology back.

Perhaps I should post AMD's word when they when with a unified GPU architecture...or is that suddenly "irrelevant" now? :hmm:
 

Skurge

Diamond Member
Aug 17, 2009
5,195
1
71
Lets see what AMD has to say:

1. A beast called the tessellator has been added which enables games developers to create smoother, less blocky and more organic looking objects in games. This is the change you’ll probably be most aware of. And it’ll show up when you look at the silhouettes of hills and mountains or the profiles of characters in games. Where artists previously had to trade off quality for performance, now artists will have the freedom to create naturalistic scenery. We’ve gotten used to seeing strangely blocky ears and noses on our opponents. But the new generation of games should allow those opponents to scare the heck out of us instead. The tessellator represents a natural next step in gaming hardware (in fact the Xbox 360 graphics chip that AMD designed already has a tessellator, and AMD graphics hardware has featured tessellator technology starting with the ATI RadeonTM HD 2000 series right up to the latest ATI RadeonTM HD 4000 series cards today).

That was pre-FERMI.

Before they got beaten in tesselation.

What you are suggestion is just the defeated trying to justify the performance deficit of their fixed function tesselation engine.

Like I said:
Hillarious.




And?
If you got the performance (NVIDIA's unified solution) it still beats the crap out of AMD's fixed function solutions

So you are advocating the slower solution...talk about holding technology back.

Perhaps I should post AMD's word when they when with a unified GPU architecture...or is that suddenly "irrelevant" now? :hmm:

Show me where AMD's tessellation hardware is insufficient?
 

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
And yet AMD keeps helping Devs Add tessellation to games. :hmm:

Yes, you go ahead and listen to PR. I'll look at real world benchmarks. Even unigene that isn't even a game with normal tessellation which is more than any game out there and AMDs cards are just as fast as nvidia's. So it doesn't matter to me who has more tessellation power if its not going to be used.

So yeah, you keep listening to all the PR both companies spew out. I think I remember nVidia saying DX11 didn't matter when they had no DX11 hardware.

Yeah, go ahead and run 64xAA or 128xAF and see what happens.

Quite funny to see people defending lower performance/qualty.

I am just waiting for you to post about how AMD is more "efficient per Watt" doing tesselation and the joke is complete ;)


This manner of posting is counter-productive and inflammatory. Please do not post in this manner.

This is a technical forum, not elementary school, and you are expected to conduct yourself accordingly.

Please familiarize yourself with the AnandTech Forum Guidelines:
We want to give all our members as much freedom as possible while maintaining an environment that encourages productive discussion. It is our desire to encourage our members to share their knowledge and experiences in order to benefit the rest of the community, while also providing a place for people to come and just hang out.

We also intend to encourage respect and responsibility among members in order to maintain order and civility. Our social forums will have a relaxed atmosphere, but other forums will be expected to remain on-topic and posts should be helpful, relevant and professional.

We ask for respect and common decency towards your fellow forum members.

(I'm going to keep quoting this same body of text, over and over again, because some of our VC&G forum members appear to have a real difficult time remembering it)

Moderator Idontcare
 
Last edited by a moderator:

badb0y

Diamond Member
Feb 22, 2010
4,015
30
91
Quite funny to see people defending lower performance/qualty.

I am just waiting for you to post about how AMD is more "efficient per Watt" doing tesselation and the joke is complete ;)

I don't get what you are arguing lol.
 

itsmydamnation

Diamond Member
Feb 6, 2011
3,045
3,835
136
Lets see what AMD has to say:
what was R600 and Xenos max tessalation factor? kinda funny how it matches exactly what they are saying now. So AMD have been consistent for what 6 generations of GPU ( Xenos R600, RV6700, RV770, everygreen, caymen) yet some how you think they have changed there tune?

fanboy will think what fanboy wants to think.

what does fixed function tellsation mean? because last time i checked you have triangle generation, you have hull and domain shaders.

And?
If you got the performance (NVIDIA's unified solution) it still beats the crap out of AMD's fixed function solutions

So you are advocating the slower solution...talk about holding technology back.

Perhaps I should post AMD's word when they when with a unified GPU architecture...or is that suddenly "irrelevant" now? :hmm:

please get a clue what your talking about, both AMD and NV have rops that work on 4 pixel quads. how about this, you can ramp up your tesselation, i will use a tesselation factor where i cant see the difference and run a higher level of AA because my rops aren't getting smashed.

k thax bye


This manner of posting is counter-productive and inflammatory. Please do not post in this manner.

This is a technical forum, not elementary school, and you are expected to conduct yourself accordingly.

Please familiarize yourself with the AnandTech Forum Guidelines:
We want to give all our members as much freedom as possible while maintaining an environment that encourages productive discussion. It is our desire to encourage our members to share their knowledge and experiences in order to benefit the rest of the community, while also providing a place for people to come and just hang out.

We also intend to encourage respect and responsibility among members in order to maintain order and civility. Our social forums will have a relaxed atmosphere, but other forums will be expected to remain on-topic and posts should be helpful, relevant and professional.

We ask for respect and common decency towards your fellow forum members.

(I'm going to keep quoting this same body of text, over and over again, because some of our VC&G forum members appear to have a real difficult time remembering it)

Moderator Idontcare
 
Last edited by a moderator:

wahdangun

Golden Member
Feb 3, 2011
1,007
148
106
I suggest a "challenge" to you.

Compare the destruactable architechture in to the scripted "destructable" architechture in BF:BC2 and tell me again you see no "difference"

(hint: Ever single hole in a wall in BF:BC" is the SAME...NO variations...as it is scripted.)

And I can only laugh at your stab at NVIDIA's unified tesselation eninge compared to AMD's limitied fixed function tesselation:

35205.png


Where does NVIDIA's tesselation implementation look unusable? :thumbsdown:

so where is the physix game that is fully destructible ? i still can't burn down the tunnel in metro, hell i even can't destroy any building/door in Batman : AA
what i have is lousy cloth and paper flying around or some smoke in metro its sooo limited i even don't care anymore, but with BF:BC2 i can do more and improvise like when i know an enemy in the building i can just destroy the wall and rush in or i will simply plant several C4 around the building and destroy it. can i do that in batman : AA ? heck i even try to put some explosive on the door but nothing happen not even a scratch, its look like the door was made from adamantium :hmm:

so i ask you what is more "realistic" ?

(Hint : i don't care how the bullet hole look like, i won't stare bullet hole while dodging enemy bullet )


i never said nv tessellation is stupid, maybe its my fault not clear enough, i mean their implementation like in metro is stupid, its just on and off ??? you said tessellation is dynamic but its look like its not dynamic at all, i mean what happen if you just have GTX 460 and want some little tessellation ??? just think what if developer make a game that set AA setting to 8 AA ??? and you can only turn it on or off isn't something like that is stupid or what ??
 

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
Again, I don't care where the physx are run, I even bought an nVidia card to try it out FFS. Its poorly done, it brings nothing to the table. The numbers behind it must be huge, but it doesnt translate to anything useful while playing the game. I don't have a problem with GPU physics, I have a problem with the GPU accelerated physx's poor execution. It does't look or perform and make the game anymore fun than CPU accelerated physx, but I need more hardware to use it? Screw that.

I agree with the need for more over-all content but don't agree with just blanketing improved Physics with just improving game-play.

There are two components in my mind-set: Fidelity and improving game-play. There is virtually no way that a developer will shoot themselves in the foot by using hardware GPU physics for dramatic game-play changes or, absolute need to have the feature by locking themselves to only one IHV in my mind. Until there are standards this may slowly become a reality but doesn't mean gamers can't have improved fidelity and modest game changing while this is in a proprietary stage.

While some offer "screw that" others may embrace choice because what other choice is there? Waiting? Isn't that fun? Who cares? Only the CPU choice for me. Does anyone really understand the resources it takes to do this? That's the key; it's not really proprietary but resources. It takes great resources to bring proprietary to the table and certainly not ideal in any way. But instead of forcing waiting for the consumer nVidia is risking, spending, and trying to bring choice for their customers. Some may not like the feature and have the ability not to use them but for some of their customers it is nice to have that choice to use or not to use. nVidia may fail but they're trying and placing resources where their mouth is.

This news to me is nice to see PhysX and DirectX 11 content from the Unreal PC platform based on DirectX 11 features and GPU physX and multi-core CPU's. There is also 3d Vision, too, which brings more quality 3d stereo for the Unreal Engine, which at times was problematic.

Where slowly entering rendering walls and for me Physics is the next frontier and why there is positioning from the powerful players.
 

wahdangun

Golden Member
Feb 3, 2011
1,007
148
106
I agree with the need for more over-all content but don't agree with just blanketing improved Physics with just improving game-play.

There are two components in my mind-set: Fidelity and improving game-play. There is virtually no way that a developer will shoot themselves in the foot by using hardware GPU physics for dramatic game-play changes or, absolute need to have the feature by locking themselves to only one IHV in my mind. Until there are standards this may slowly become a reality but doesn't mean gamers can't have improved fidelity and modest game changing while this is in a proprietary stage.

While some offer "screw that" others may embrace choice because what other choice is there? Waiting? Isn't that fun? Who cares? Only the CPU choice for me. Does anyone really understand the resources it takes to do this? That's the key; it's not really proprietary but resources. It takes great resources to bring proprietary to the table and certainly not ideal in any way. But instead of forcing waiting for the consumer nVidia is risking, spending, and trying to bring choice for their customers. Some may not like the feature and have the ability not to use them but for some of their customers it is nice to have that choice to use or not to use. nVidia may fail but they're trying and placing resources where their mouth is.

This news to me is nice to see PhysX and DirectX 11 content from the Unreal PC platform based on DirectX 11 features and GPU physX and multi-core CPU's. There is also 3d Vision, too, which brings more quality 3d stereo for the Unreal Engine, which at times was problematic.

Where slowly entering rendering walls and for me Physics is the next frontier and why there is positioning from the powerful players.

yeah physics really need to be open standard, and use open platform like direct compute or openCL, and next 2 years i bet most people will have APU CPU on their computer either from AMD or intel, and it will be waste of silicon and die space if developer didn't start to using this untaped power

and no we will never be near rendering wall because next step from raster is ray-tracing
 

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
I would like to see nVidia port the GPU physic aspect to OpenCL. I'm a firm believer in using the strengths of the CPU and GPU because both have their strengths.

Would like to learn more how much mature Cuda may be over OpenCL though.

Proprietary in my mind-set does bring division, chaos and fragmentation but on the other side of the coin it brings innovation and choice, which hopefully raises awareness so standards may be forged to mature.
 

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
I would like to see nVidia port the GPU physic aspect to OpenCL. I'm a firm believer in using the strengths of the CPU and GPU because both have their strengths.

Would like to learn more how much mature Cuda may be over OpenCL though.

Proprietary in my mind-set does bring division, chaos and fragmentation but on the other side of the coin it brings innovation and choice, which hopefully raises awareness so standards may be forged to mature.


CUDA is years ahead of OpenCL, just look at the "eco-system".

People are teaching CUDA in schools...not OpenCL....because of the eco-system.

And don't forget that DirectX is proprietary and OpenGL is open...look how that went.
 

Red Storm

Lifer
Oct 2, 2005
14,233
234
106
Personally I'm still waiting for PhysX to actually beat games like Crysis or even Half-Life at actual physics that affects the game world rather than just enhanced graphics. Nothing wrong with better looking graphics, but when it's strictly visual then it has no real effect on the game play, which is what GPU accelerated physics should be all about!
 

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
I desire more. Why wouldn't a person desire physics that offer more realism, improved fidelity and redefine game-play? Getting closer and closer with the introduction of GPU Physics and multi-core CPU's, and the powerful players creating tools.
 

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
And yet AMD keeps helping Devs Add tessellation to games. :hmm:

Yes, you go ahead and listen to PR. I'll look at real world benchmarks. Even unigene that isn't even a game with normal tessellation which is more than any game out there and AMDs cards are just as fast as nvidia's. So it doesn't matter to me who has more tessellation power if its not going to be used.

So yeah, you keep listening to all the PR both companies spew out. I think I remember nVidia saying DX11 didn't matter when they had no DX11 hardware.

Yeah, go ahead and run 64xAA or 128xAF and see what happens.

Mike Hara said:
DirectX 11 by itself is not going be the defining reason to buy a new GPU. It will be one of the reasons. This is why Microsoft is in work with the industry to allow more freedom and more creativity in how you build content, which is always good, and the new features in DirectX 11 are going to allow people to do that. But that no longer is the only reason, we believe, consumers would want to invest in a GPU

It was taken out of context as DirectX 11 is not important or didn't matter.
 
Last edited:

sandorski

No Lifer
Oct 10, 1999
70,677
6,250
126
OpenCL Apps are really starting to flow right now. PhysX's days are numbered.
 

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
Wouldn't Cuda and OpenCL have more in common? Personally would like to see OpenCL flourish and be more mature so nVidia can port PhysX -- so developers may possibly offer more content than what is offered right now.
 

Lonyo

Lifer
Aug 10, 2002
21,938
6
81
I desire more. Why wouldn't a person desire physics that offer more realism, improved fidelity and redefine game-play? Getting closer and closer with the introduction of GPU Physics and multi-core CPU's, and the powerful players creating tools.

That's exactly what people want.
That's why many people don't give a damn about PhysX.
 

OCGuy

Lifer
Jul 12, 2000
27,224
37
91
It seems like nV is always trying to push innovation with the developers. AMD needs to step up for the good of PC gaming on the whole.
 

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
That's exactly what people want.
That's why many people don't give a damn about PhysX.

You don't like more realism and more fidelity? It just has to be improved game-play and that's it?

What the hell is the big deal of filtering? Anti-aliasing? Tessellation? Ambient occlusion? Shadows for the most part? Improved lighting? Who cares about realistic Physics to improve the environment, enhance the mood and bring more realism, just change and redefine game-play and that's it.

Asking for a little too much too soon, one may imagine, and one usually has to crawl, take a few steps, walk then run. Create hardware, create tools, create applications and as software and hardware mature things get much more ideal.

Some allow idealism to be the enemy of good.
 

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
OpenCL Apps are really starting to flow right now. PhysX's days are numbered.

Would you stop comparing OpenCL to PhysX?

Damn, I need to write a FAQ/Informative post...the amount of false claims is staggering :\