OMF:BG Message Board Forum Index OMF:BG Message Board
One Must Fall: Battlegrounds
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Accelerated physics article
Goto page 1, 2  Next
 
Post new topic   Reply to topic    OMF:BG Message Board Forum Index -> Tech Talk
View previous topic :: View next topic  
Author Message
NightShadow



Joined: 20 Feb 2003
Posts: 2674
Location: Sennadar

PostPosted: Tue Jul 11, 2006 8:43 am    Post subject: Accelerated physics article Reply with quote

http://enthusiast.hardocp.com/article.html?art=MTA5NywxLCxoZW50aHVzaWFzdA==

I found the article to be rather informative. Cleared up a couple misconceptions that I had.
_________________
[Elite] pilot of the Force Flying Circus
"Just because you are unique does not mean you are useful"
Back to top
View user's profile Send private message AIM Address Yahoo Messenger MSN Messenger
Discomb



Joined: 31 Jul 2003
Posts: 4262
Location: Beijing, China

PostPosted: Tue Jul 11, 2006 9:38 am    Post subject: Re: Accelerated physics article Reply with quote

Is it just me or does the NVIDIA representitive really not make a whole lot of sense? I mean, the ATI guy kept boasting about how great they were, but he was also excellent at laying the facts on the table. I walked away with very little information from the second page... :crazy:
_________________
Back to top
View user's profile Send private message Send e-mail ICQ Number
salem



Joined: 07 Aug 2003
Posts: 1778
Location: Spain

PostPosted: Tue Jul 11, 2006 11:40 am    Post subject: Re: Accelerated physics article Reply with quote

Interesting, i'll take a read Smile
Back to top
View user's profile Send private message
killer_roach



Joined: 18 Jan 2001
Posts: 8048
Location: Lexington, Kentucky

PostPosted: Tue Jul 11, 2006 11:59 am    Post subject: Re: Accelerated physics article Reply with quote

From what I read, seemed like the interview with Havok was the most interesting of all of them... ATI and nVidia's statements both seemed more like marketing than anything (although nVidia's was blatantly so).
_________________
Official forum economist. Explodes when thrown.

Desktop:
Core i7-2600K @ 4.5 GHz
16GB DDR3-1600
eVGA GeForce 970 FTW
Windows 10 Preview

Laptop:
Core 2 Duo 2.8 GHz
4GB DDR3-800
Radeon 4570 1GB
Windows 7 Home Premium x64
Back to top
View user's profile Send private message AIM Address Yahoo Messenger MSN Messenger ICQ Number
Discomb



Joined: 31 Jul 2003
Posts: 4262
Location: Beijing, China

PostPosted: Tue Jul 11, 2006 2:37 pm    Post subject: Re: Accelerated physics article Reply with quote

I think the ATI's statement was, though advertising in nature, an excellent summary of most of the things that would be touched later on by the other interviewees, so I'll give him credit for being so good at setting things straight.

The second article told me pretty much nothing. It was confusing and uninteresting.

The third (Havok) was probably the most insightful and attacked the problem from quite a few angles, demonstrating the depth to which all these things are being considered at the moment.

And finally, the PhysX article had some excellent examples and was excellent at making things quite simple (despite being a half advertisement on its' own), but after reading through three pages of the same thing, no one honestly cares for a summary of what the difference between game and effect physics are.

The only thing I found odd is that while NVIDIA and ATI both claimed that there is no difference between the two physics types, Havok (I think it was) said that they could reach 10x the complexity with effect physics, since apparently if the player doesn't have to interact with them, it lowers the calculation burden. I'm not entirely sure how that works, but Havok does a lot of things through the [filtered] from what I know, so heh.

So the PhysX marketing strategy is "Uber CPU and GPU setups are great, but you want to buy our [filtered], because we do it better and faster, freeing up resources." I think there's a lot of fundamentally wrong things with this strategy, but heh, this whole post was rather pointless. Just writing for myself, to make sure I understand everything that's just been said.
_________________
Back to top
View user's profile Send private message Send e-mail ICQ Number
Endy



Joined: 31 Jul 2003
Posts: 4079
Location: Lost within my own thoughts.

PostPosted: Tue Jul 11, 2006 4:42 pm    Post subject: Re: Accelerated physics article Reply with quote

This interview was very informative, apparently both NVidia and ATI are the single greatest corporations in the history of mankind, also there are physics involved in this.

Inverviewer: So I hear your company is gearing up to do physics acceleration in your next gen cards?
Interviewee: Yeah, we are pretty awsome.
Interviewer: Can you give us any insights into what exactly your plan is to bring physics to the market?
Interviewee: Our physics will be approximatley 9999999999.99% faster than the offerings of all our competitors at release.
Interviewer: What?
Interviewee: Someday, after we evolve into beings of pure energy we will conquer the universe with our mere thoughts. Did I mention we are awsome and also have physics!!

...and so on. :thumbsdown: :thumbsdown: :thumbsdown:
_________________
"I reject your reality, and substitute my own!" - Adam Savage (MythBusters)
"I don't know the key to success, but the key to failure is trying to please everyone." - Bill Cosby
Back to top
View user's profile Send private message
Rob Elam



Joined: 18 Sep 1998
Posts: 2122
Location: Panama City

PostPosted: Tue Jul 11, 2006 9:32 pm    Post subject: Re: Accelerated physics article Reply with quote

Quote:
The only thing I found odd is that while NVIDIA and ATI both claimed that there is no difference between the two physics types, Havok (I think it was) said that they could reach 10x the complexity with effect physics, since apparently if the player doesn't have to interact with them, it lowers the calculation burden. I'm not entirely sure how that works, but Havok does a lot of things through the [filtered] from what I know, so heh.

I think the point was that the GPU solution can do everything that the PhysX solution can, which is true. Basically, PhysX is pushing the idea that it can do “gameplay” physics, and conversely GPU solutions cannot. In reality, GPU solutions can, they just don’t in the very first implementation of HavoxFX, which is retrofitted on existing hardware and is not intended to do “gameplay” physics. From a purely architectural standpoint (IOW, disregarding what may not be fully implemented in software APIs) there’s no difference between the PhysX solution and GPU solutions, except GPU solutions can also do graphics and can do effects physics (physics that are calculated and drawn without significant CPU interaction). GPU solutions can do what PhysX does, and can do what HavokFX does, but many have argued that GPU solutions can ONLY do what HavokFX does, which is incorrect. In the end, the GPU solution is architecturally superior, and can do anything that PhysX can do, which is likely the misconception ATI/Nvidia will strive to dispel. IMO, this won’t be hard, PhysX is dead, deader than disco, and without divine intervention I can’t see how it can go any other way.
Back to top
View user's profile Send private message
Discomb



Joined: 31 Jul 2003
Posts: 4262
Location: Beijing, China

PostPosted: Wed Jul 12, 2006 2:23 am    Post subject: Re: Accelerated physics article Reply with quote

My question was essentially in that the first two articles said that there is no difference between effect and game physics, while the other two said there is a big difference in calculation burden. I'm not sure if you meant to answer that or not but you certainly did anyway. :p
_________________
Back to top
View user's profile Send private message Send e-mail ICQ Number
NightShadow



Joined: 20 Feb 2003
Posts: 2674
Location: Sennadar

PostPosted: Wed Jul 12, 2006 11:15 am    Post subject: Re: Accelerated physics article Reply with quote

Yeah, I was under the impression that video cards couldn't currently do gameplay physics if they wanted to. I knew they'd be able to eventually, I thought it wouldn't be until the physics drivers matured a bit or something. Smile

PhysX may die out, but I don't think dedicated physics solutions will. PhysX is just too early; there's no need of a separate physics card if only effects will make use of it, and if video cards can do the same thing without sacrificing anything. Once gameplay physics are relied on heavily, and video cards start slowing down due to physics processing, a separate physics card could become successful. It also can't rely on propietary physics stuff like Ageia; it should all be done through DirectX or whatever.
_________________
[Elite] pilot of the Force Flying Circus
"Just because you are unique does not mean you are useful"
Back to top
View user's profile Send private message AIM Address Yahoo Messenger MSN Messenger
Rob Elam



Joined: 18 Sep 1998
Posts: 2122
Location: Panama City

PostPosted: Wed Jul 12, 2006 12:21 pm    Post subject: Re: Accelerated physics article Reply with quote

Quote:
Yeah, I was under the impression that video cards couldn't currently do gameplay physics if they wanted to.

Technically they can, but you can’t grab data from video cards fast enough for it to be worthwhile (they have been designed to stream data to the card fast, not the other way around). That’s the part that they have/will fix (I have no idea how much of the problem is hardware or software).
Quote:
PhysX may die out, but I don't think dedicated physics solutions will.

No, they won’t die out, because they have to live before they can die Smile. They’ll never get off the ground (perhaps at specialized levels, but not consumer levels). If the GPUs get overloaded, they’ll just move to multi-core solutions (or more cores than they have now), the best time in the market for a dedicated physics solution is now, and it isn’t that great. It will not get any easier than this since GPU solutions will become more prominent.
Back to top
View user's profile Send private message
killer_roach



Joined: 18 Jan 2001
Posts: 8048
Location: Lexington, Kentucky

PostPosted: Wed Jul 12, 2006 12:33 pm    Post subject: Re: Accelerated physics article Reply with quote

This is where AMD's 4x4 initiative will probably come into play... imagine dropping a GPU into a motherboard socket, with dedicated DDR3 or DDR4 and a HyperTransport 3.0 link to the CPU... and then your typical GPU in a PEG slot Smile
_________________
Official forum economist. Explodes when thrown.

Desktop:
Core i7-2600K @ 4.5 GHz
16GB DDR3-1600
eVGA GeForce 970 FTW
Windows 10 Preview

Laptop:
Core 2 Duo 2.8 GHz
4GB DDR3-800
Radeon 4570 1GB
Windows 7 Home Premium x64
Back to top
View user's profile Send private message AIM Address Yahoo Messenger MSN Messenger ICQ Number
Raven



Joined: 14 Oct 2000
Posts: 6363
Location: Sriniskal

PostPosted: Wed Jul 12, 2006 3:27 pm    Post subject: Re: Accelerated physics article Reply with quote

I really wish games would just decentralize the **** out of everything and let us decide for ourselves what's going to power the game itself, the physics, and graphics.

I don't know where I'm going with this, but with dual core processors, SLI, and eventually quad core processors and dual core video cards, you should be able to get by without an extra processor just for physics. A game should be able to simply go "Ok, you've got a 2 ghz dual core and 1 video card, so physics and triggers go to CPU 2, AI and everything else on CPU 1, and graphics on GPU 1"
_________________
"The unofficial first person to edit their sig since 2009! Possibly 2008."

Scorpion Navy Issue
Back to top
View user's profile Send private message AIM Address ICQ Number
Vuen



Joined: 31 Aug 1999
Posts: 4968
Location: Canada

PostPosted: Wed Jul 12, 2006 8:08 pm    Post subject: Re: Accelerated physics article Reply with quote

The ATI representative was an idiot. He said absolutely nothing useful:

Quote:
“Gameplay” physics can also improve the look and feel of a game, but these physics may have a direct impact on the way a game plays. For example, your character’s weight within a given environment, or whether he simply falls through the floor or actually walks on it are both examples of basic gameplay physics.


That's not taken out of context; that's his introduction to "Gameplay physics."

The NVIDIA interview on the other hand was very informative. He actually explained why we have this distinction between gameplay and effects. He provided a long list of advantages to dealing with effects physics and introduced real issues with gameplay physics; for example, in a multiplayer game, all clients need to produce the same physical result to stay synchronized, so you can't just turn it off if your system isn't powerful enough.

I kinda gave up reading half way through the HavokFX interview. In any case, I just don't see that big of a need for gameplay physics to advance much beyond what we have today before they can start a revolution in gameplay. We've seen the likes of what the gravity gun can do in HL2, so why aren't there already a swarm of games that are taking advantage of gameplay physics? Games where you can shoot pop cans off of window sills? Come on. Personally I think developers just don't have any ideas. They have no idea what they can accomplish with physics simulations other than "look at this shiny debris bouncing around", so they're only pushing effects physics. This is what ten years of making [filtered] FPS games will do to a developer.



Quote:
Technically they can, but you can’t grab data from video cards fast enough for it to be worthwhile (they have been designed to stream data to the card fast, not the other way around). That’s the part that they have/will fix (I have no idea how much of the problem is hardware or software).


The NVIDIA rep had this to say:

Quote:
Simulation of gameplay objects would increase the amount of data read back to system memory, over and above what’s currently being transferred back. Yet, these amounts of data are very small when compared to total PCI-E readback bandwidth, so we wouldn’t anticipate a performance hit.



To relate this to the console discussion, I find it ironic that physics isn't what the PS3 is pushing for their next-gen console. Developers today are saying that they aren't implementing gameplay physics because the dedicated hardware isn't there. A console is the perfect solution to this problem. So, where are they?
Back to top
View user's profile Send private message
NightShadow



Joined: 20 Feb 2003
Posts: 2674
Location: Sennadar

PostPosted: Wed Jul 12, 2006 8:49 pm    Post subject: Re: Accelerated physics article Reply with quote

Thanks, Vuen. I was beginning to think I was the only one who thought ATI's response was dumb.

And, speaking of console physics, anybody hear anything new about the rumoured PPU the Wii is supposed to have? http://www.codenamerevolution.com/?p=776
_________________
[Elite] pilot of the Force Flying Circus
"Just because you are unique does not mean you are useful"
Back to top
View user's profile Send private message AIM Address Yahoo Messenger MSN Messenger
Rob Elam



Joined: 18 Sep 1998
Posts: 2122
Location: Panama City

PostPosted: Wed Jul 12, 2006 9:23 pm    Post subject: Re: Accelerated physics article Reply with quote

Quote:
And, speaking of console physics, anybody hear anything new about the rumoured PPU the Wii is supposed to have? http://www.codenamerevolution.com/?p=776

The safe assumption is that it's not a seperate PPU, but the ATI solution discussed here (remember, they've got a newer ATI core).
Back to top
View user's profile Send private message
Display posts from previous:   
Post new topic   Reply to topic    OMF:BG Message Board Forum Index -> Tech Talk All times are GMT - 8 Hours
Goto page 1, 2  Next
Page 1 of 2

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group

Style Blacker1 by Blacker47