PDA

View Full Version : PhysX - Novodex physic engine - BoB



Zoom2136
03-22-2006, 08:33 AM
Will BoB support such a chip....and physic engine http://forums.ubi.com/images/smilies/touche.gif

Zoom2136
03-22-2006, 08:33 AM
Will BoB support such a chip....and physic engine http://forums.ubi.com/images/smilies/touche.gif

AKA_TAGERT
03-22-2006, 08:46 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Zoom2136:
Will BoB support such a chip....and physic engine http://forums.ubi.com/images/smilies/touche.gif </div></BLOCKQUOTE>Doubt it, dedicated physic CPUs are just not that great for FM calculations. They might be useful in doing other things like bullet trajectorys and stuff, but, I dont see these dedicated physic CPUs making a bigger splash than MMX did a few years back. In that about the time you get them coded up, a new Intel or AMD processor comes out that is fast enough to do all those calcs in less time the old fashion way.

PS your a little late
http://forums.ubi.com/eve/forums/a/tpc/f/63110913/m/8071063524
http://forums.ubi.com/eve/forums/a/tpc/f/63110913/m/1551050524

ManicGibber
03-22-2006, 10:04 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by AKA_TAGERT:
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Zoom2136:
Will BoB support such a chip....and physic engine http://forums.ubi.com/images/smilies/touche.gif </div></BLOCKQUOTE>Doubt it, dedicated physic CPUs are just not that great for FM calculations. They might be useful in doing other things like bullet trajectorys and stuff, but, I dont see these dedicated physic CPUs making a bigger splash than MMX did a few years back. In that about the time you get them coded up, a new Intel or AMD processor comes out that is fast enough to do all those calcs in less time the old fashion way.

PS your a little late
http://forums.ubi.com/eve/forums/a/tpc/f/63110913/m/8071063524
http://forums.ubi.com/eve/forums/a/tpc/f/63110913/m/1551050524 </div></BLOCKQUOTE>

Actually your the one that is late http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif<A HREF="http://forums.ubi.com/eve/forums/a/tpc/f/63110913/m/2441091113/r/2441091113#2441091113" TARGET=_blank>
http://forums.ubi.com/eve/forums/a/tpc/f/63110913/m...113#2441091113 (http://forums.ubi.com/eve/forums/a/tpc/f/63110913/m/2441091113/r/2441091113#2441091113%5B/URL%5D)</A>

Jaws2002
03-22-2006, 11:54 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by AKA_TAGERT:
Doubt it, dedicated physic CPUs are just not that great for FM calculations. They might be useful in doing other things like bullet trajectorys and stuff, but, I dont see these dedicated physic CPUs making a bigger splash than MMX did a few years back. In that about the time you get them coded up, a new Intel or AMD processor comes out that is fast enough to do all those calcs in less time the old fashion way.

</div></BLOCKQUOTE>



I think it would help flight sims more then other games. Think of how much load it can take off the processor or GPU: Damage model, ballistics, clouds, fog, mist, water waves.

Flight model in general since flying is nothing but the interaction between aircraft and air.

Tator_Totts
03-22-2006, 12:53 PM
I like the graphic sli physic chip.

So, how much faster is SLI Physics versus traditional CPU physics? We have to wait to get the driver - and supporting applications - into our hands to be able to judge the capability of the technology. Nvidia gave out some preliminary figures achieved in a demo that shows 15,000 boulders colliding with each other. The demo, aptly nicknamed "BoulderMark" was run by an SLI Physics enabled machine and GeForce 7900 GTX graphics cards with 64.5 frames per second while the CPU-only system topped out at out 6.2 frames per second.

http://www.tgdaily.com/2006/03/20/nvidia_sli_forphysics/

NonWonderDog
03-22-2006, 01:04 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Jaws2002:
I think it would help flight sims more then other games. Think of how much load it can take off the processor or GPU: Damage model, ballistics, clouds, fog, mist, water waves.

Flight model in general since flying is nothing but the interaction between aircraft and air. </div></BLOCKQUOTE>

Clouds and munitions probably would be the best implementation of PhysX in a sim. Flight model calculations just don't work that way. While flying is indeed nothing more than the interaction betwen the aircraft and the air, the flight models are based on the general equations you learn at university, not computational fluid dynamics. You would need a LOT more than 50,000 particles to do CFD on sky full of aircraft in real time.

Swirling clouds, though, would be neat.

iroseland
03-22-2006, 01:56 PM
Hey,

Just getting support to pretty up things like clouds, fire, ballistics, the damadge model, fog and mist and wht not alone would be great. Part of the problem is that the cpu's are kind of not all that good at preforming lots and lots of floating point. Anything that can or could be offloaded to a card that can do it faster would result in more spare cycles to handle calculations for the flight model.

What really needs to happen to make these take off is a game that make people want to have them. Even with fantastic PPU intigration BoB would probably not be that game. Though that would be nice. 8^)


BSS_AIJO

Codex1971
03-24-2006, 05:05 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by NonWonderDog:
Flight model calculations just don't work that way. While flying is indeed nothing more than the interaction betwen the aircraft and the air, the flight models are based on the general equations you learn at university, not computational fluid dynamics. </div></BLOCKQUOTE>

True...but remember these physics chips are nothing more than high speed number crunchers. A dedicated physics chip would allow much more complex calculations to be made. You don't really need to use particles to simulate air flow, you could generalise those particles into something like ribbons or bands which can represent air flow. The physics chip can then manipulate those ribbons by using more complex aerodynamic formula's (like in fluid dynamics) which would, in turn, give a more accurate FM. And doing it in real time.

http://wwwcg.in.tum.de/Research/Projects/Windtunnel

Obviously a dual CPU PC with a physics chip (or two) could not produce calculations to what is shown in the above link, but it could certainly do the same type of calculations using a lower sample resolution.

I equate using a dedicated physics chip would be like going from paper calculations to wind tunnel testing.

NonWonderDog
03-24-2006, 09:15 AM
Yeah, but if you want to do that, there's really no reason to do it in real time. You could just find lift and drag curves beforehand and program them in... just like the sim does now. This is probably EXACTLY how the Go-229 and Bf-109Z flight models came into being, ya know.

I just don't see any *reason* to use any kind of real-time CFD for flight models unless you have the millions of particles needed to find forces on any arbitrary orientation of the plane.

The only thing I'm not sure about is X-Plane. Does X-Plane actually do anything in real-time? They claim to use blade-element analysis for everything, but I still don't see how they could use blade-element on anything but the wings and prop. PhysX might eventually be a good thing for X-Plane flight models, but I don't think it's a good idea to do what they do for a combat flight simulator with a pre-set plane list.

Even if X-Plane comes up with some way to compute FM in real time, will it be good enough for the Ubiwhiners? I'm not sure it's even possible to get documented performance by using a low-res CFD on computer game models. People already deride the Gotha and 109Z for having "fantasy" flight models... what if every plane had a (real-time) "fantasy" flight model? Would anyone accept that the flight models are correct because they're calculated? What if they don't fly like pilot accounts say?

I'm still exited for the PhysX stuff because of what it might do for clouds and explosions, but I don't expect it to have any direct effect on flight models.

Codex1971
03-24-2006, 05:40 PM
Good point NonWonderDog... http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif

However I just can't see why a dedicated physics chip would only be good for just bouncing barrels and bodies all over the screen in a FPS. http://forums.ubi.com/images/smilies/35.gif

As you said it would make weather effects and explosions interesting but surely it could also be put to good use in enhancing a FM in some way...maybe damage modeling, prop wash from bombers, turbulance etc.

As for pilot accounts...since when has that been taken into consideration in the sim anyway... http://forums.ubi.com/images/smilies/16x16_smiley-very-happy.gif

BueJack
03-25-2006, 07:43 AM
AS you say.. I think IL2 FM is a batch of numbers in a LUT (look-up-table). A lot of number crunching is done on the GPU card (put in a 'lesser' GPU card and watch Frame rate sink).

The Physics card will offload a lot of the number crunching from the CPU/GPU, and as mentioned help with the finer effects of a sim.

I'd like to see the physics chip standardised on a mobo http://forums.ubi.com/groupee_common/emoticons/icon_biggrin.gif

AKA_TAGERT
03-25-2006, 08:10 AM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by BueJack:
AS you say.. I think IL2 FM is a batch of numbers in a LUT (look-up-table). </div></BLOCKQUOTE>Nope, last sim made like that was back in th mid 90s. Since then they have all been 6DOF FMs

BueJack
03-25-2006, 09:46 AM
This is the way I think of it...

One has a generic flight model (3DOF or 6DOF is irrelevant) as there are so many planes in the game. Each plane's FM comes with a few 'thousand' parameters in the form of a LUT.

Each plane in the scnario has it's FM LUT plugged into the Generic FM.

Makes sense to me... http://forums.ubi.com/images/smilies/25.gif

AKA_TAGERT
03-25-2006, 08:31 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by BueJack:
This is the way I think of it...

One has a generic flight model (3DOF or 6DOF is irrelevant) as there are so many planes in the game. Each plane's FM comes with a few 'thousand' parameters in the form of a LUT.

Each plane in the scnario has it's FM LUT plugged into the Generic FM.

Makes sense to me... http://forums.ubi.com/images/smilies/25.gif </div></BLOCKQUOTE>You are corret! I thought you were saying it was a table based FM where no real time calcualtions were being performed and all results were just part of a lookup table.. Which is how they use to do it back in the 90s on a PC (aka 386 or 286).

As you hinted at, there is only one FM! That is to say the math does not change, only the value you load into it. For example

f(x) = A*x^2 + 1

Now when simulating a P38, A might be equal 100 and when simulating a Fw190 A might be equal to 111. The values of A are in a table that get loaded into the math of the equation once you select a plane type. But the math itself does not change.

Keep in mind, there are no absolute rules here, Im sure there may be a few versions of FM math, like single or twin engine and etc. But the basic math is the same.