PDA

View Full Version : Agiea PhysX



FCIProject
07-17-2005, 01:09 PM
Are 1C:Maddox Games going to support the Agiea PhysX Physics Processing Unit (PPU) in their prodicts?

It woild be a great benefit if IL2 and Battle of Britain supported the PPU. This would mean a huge amount of pyhsics processing would be removed from the CPU. This would free up the CPU for other tasks such as taking the scenary to a new level or a more complex Dynamic Campaign Engine.

I look forward to your repsone Oleg.

FCIProject
07-17-2005, 01:09 PM
Are 1C:Maddox Games going to support the Agiea PhysX Physics Processing Unit (PPU) in their prodicts?

It woild be a great benefit if IL2 and Battle of Britain supported the PPU. This would mean a huge amount of pyhsics processing would be removed from the CPU. This would free up the CPU for other tasks such as taking the scenary to a new level or a more complex Dynamic Campaign Engine.

I look forward to your repsone Oleg.

Bearcat99
07-17-2005, 05:32 PM
I cant imagine 1C not considering this technology... actually IMO it works out great because it may delay BoB long enough for the technology to catch up and give 1C a legitimate reason to continue to support FB in the interim. I know most of us would gladly pay for add ons...... I know I would.

FCIProject
07-17-2005, 05:57 PM
IMHO I want 1C to move onto Battle of Britain. I would like to see the bigger map in action, which would mean having to navigate more accurately than is currently required.

With PPU it would be possible to have larger maps and more accurate scenery. This could be extended to other maps, not just the south of England.

I hope 1C are also looking at 64bit XP. The 32bit XP 2Gb memory limit would be a thing of the past. Allowing for more map, AI and campaign data to be held in memory.

darkhorizon11
07-22-2005, 12:03 PM
Bump to this! Flight Physics is the biggest weakness to the IL2 series right now...

FCIProject
07-22-2005, 12:37 PM
IMO I don't believe the physics is the weakest part of IL2. The lack of a true Dynnamic Campaign is. IMHO, the weakest part of IL2.

However, with a PPU calculating the physics shouldn't this reduce the argument of the physics for planes being incorrect or not?

As long as the PPU is producing real-world physics shouldn't the planes 'fly right'?

I have looked at the Agiea demos and the physics 'look right', but will this be translated to how a plane flies or a racing car drives, etc...

If 1C used the PPU they could concerntrate on very detailed aircraft models for damage modelling, etc. Thet could also take the scenery engine to a different level, with more ground detail, buildings, rivers and the like. All with damage models.

The much wanted Dynamic Campaign engine could also be developed, in house.

Re-inventing the Physics Engine 'wheel' should be a thing of the past with the PPU.

lbhskier37
07-22-2005, 12:43 PM
PPU is a partical physics processor that is great for making real looking explosions in FPSs, but not really designed for fluid dynamics. In order for BoB to have a flight model that went to a hardware physics processor, everyone would have to have one, and I don't think everyone is going to have a hardware physics processor just yet. The PPU will probably get supported in the FPS market because all it does is make pretty explosions and falling stuff, graphic stuff that can be turned off in a game if you don't have the hardware to support it, but its not like you can turn off the flightmodel if you don't have it so that would eliminate anyone who hasn't bought one. What I see as more likely is them taking advantage of dual cores, since looking at roadmaps shows everything becoming dual core by 2006, so by 2007 when BoB would probably come out, dualcore would be mainstream. Going with a flight model that takes good advantages of dual cores would also work great because nothing would prevent it from working if you only had a single core, you would just have a performance hit. In addition using the CPU allows much more flexibility, using a physics processor you are limited to what is hard coded in the processor, limiting advancements to more complex flight models as the game matures (as we have seen in IL2). But maybe they will support it to make explosions look prettier http://forums.ubi.com/groupee_common/emoticons/icon_smile.gif

lbhskier37
07-22-2005, 12:51 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by FCIProject:

As long as the PPU is producing real-world physics shouldn't the planes 'fly right'?

I have looked at the Agiea demos and the physics 'look right', but will this be translated to how a plane flies or a racing car drives, etc...

</div></BLOCKQUOTE>

Some card will never in the near future produce "real-world physics" it is always going to be an approximation. Its pretty easy to make flying bricks "look right", it doesn't mean they are actually flying right. There are multi-thousand dollar programs that run on some of the most advanced hardware out there that is used to create "virtual" wind tunnels. This hardware and software is getting pretty good at creating good fluid dynamics representations, but its nowhere near realtime yet even on the thousands of dollars worth of hardware it takes. The PPU is probably **** good at making assumptions to make some exploding debris look cool, but since it's being marketed towards FPSs and not our tiny little niche market I'm sure they haven't concentrated on proper fluid dynamics.

FCIProject
07-22-2005, 01:06 PM
Some good points there. The main one being not everybody wanting to buy a PPU. Asus are going to be one of the first to have a PPU enabled board.

If the PPU is supported by Flight Sims, like BoB or, dare I say, a new version of MS Flight Sim I would buy one. How much do we spend on graphics cards and PCs, just for IL2, now?

The other point regards Fluid Dynamics. This is a huge nut for a PC to crack. No game has Fluid dynamics in it now. Why does not modelling Fluid Dynamics have to stop a PPU being used?

I have not seen the Agiea chip is only FPS.

The more hardware that can be used to offload CPU usage the better for all types of games, including simulations.

Bearcat99
07-22-2005, 01:28 PM
I gaurantee you that in 5 years every new mobo will have a ppu byuilt into it.. remember whe onboard sound was new... and hokey? Now it is better than some soundcards.. I use Soundstorm which I think is great.

lbhskier37
07-22-2005, 01:45 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by FCIProject:

The other point regards Fluid Dynamics. This is a huge nut for a PC to crack. No game has Fluid dynamics in it now. Why does not modelling Fluid Dynamics have to stop a PPU being used?

I have not seen the Agiea chip is only FPS.

</div></BLOCKQUOTE>

Any game that doesn't use tables to calculate flight models uses some sort of fluid dynamics. Actually fluid dynamics is really just a simplification of conservation of mass, conservation of momentum and conservation of energy. The physics of anything in motion is a simplification of these things. I haven't seen anything stating for sure the Agiea chip couldn't do a flight model, but I haven't seen anything showing it to be more than just something that can calculate the projectile motion of lots of little particles (great for exploding stuff, useless for flight dynamics).

About integrating it, sound is integrated because nearly everyone can use it, from the gamer down to the corporate stiff, but that guy writing TPS reports doesn't have any use for a physics processor. Physics chips will probably be stuck in high end gaming boards but thats about it. Maybe in 5-10 years there will be a physics chip advanced enough to really replicate physics, but I think it will come more in the form of one of the cores of a multicore processor being specifically designed to do physics calculations.

FCIProject
07-23-2005, 08:05 AM
That puts me in my place regarding Fluid Dynamics
http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif

If Agiea want to have a successful product then it must do more than pretty explosions and only support FPS.

The PPU should work for all types of games. I am not talking just Flight Sims.

In the long run CPUs will have dedicated physics extensions, such as MMX and all the other stuff that has been added on over the years. This is the just the begining of dedicated physics calculations.

It will be interesting to see the first (non-FPS) game using a PPU.

Wolf-Strike
07-24-2005, 10:05 AM
I would think that this PPU will be obsolete since we are now moving into the dual core processors.

Now one core for physics and the other for the rest.Just needs the program to run correctley.From what I read the slowdown in processor speeds advancing is that they just cant physically make the lines of transmission any smaller.So that why they are now moving to two cores on one chip to counter this effect until a more advanced way of making semi conducters allows them to go even smaller.

Targ
07-24-2005, 11:38 AM
I would think the duel core chips and quad chips in the near future will have more relavence than the PPU physics chip in the future.
It will be tough to buy a computer in 2007 that does not have a duel core cpu.

FCIProject
07-24-2005, 05:36 PM
Dual core chips are main stream already, however not many applications (games) are multi-threaded and can't take advantage of miltiple CPUs.

The reason Intel went dual core is the heat generated by P4 chips. When Intel announced the P4 they said it would eventually run at 5Ghz. It was never going to get to that speed without going nuclear. Intel had to re-write its roadmap, to compete, so came up with dual core.

AMD followed because of marketing pressure. Until applications are multi-threaded then dual core isn't going to be a distinct performance boost.

The PPU is like a graphics card; it offloads a particular aspect of processing from the CPU.

How can the PPU be obsolete? It hasn't been given the chance to prove itself, one way or the other.

Wolf-Strike
07-24-2005, 10:03 PM
If we are gonna have two 3.0 Ghz processors than run one for physics and thats why the PPU will be obsolete.Its just my opinion.

FCIProject
07-25-2005, 12:45 AM
Your opinion is very welcome Wolf-Strike.

Saying the physics could run on one CPU would still mean writing a multi-threaded application.

It may be easier to write the physics engine for the second processor, because a high-level language like C++ can be used.

However, wouldn't it be just as easy to write API calls to the PPU and let that do the physics, without re-inventing the wheel?

Of course, I am not a games writer, so this is a over simplification of the necessary approach.

I do have bit of experience writing multi-threaded code though and it ain't easy.

csThor
07-25-2005, 11:13 PM
<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">However, wouldn't it be just as easy to write API calls to the PPU and let that do the physics, without re-inventing the wheel? </div></BLOCKQUOTE>

There's one thing you forget - Dual-Core CPU's might become a widespread standard for modern PCs (I think they will), but a PPU will always remain a niche product. If you link your software to such a device - a software that is a niche product, too - you're in for low sales as it scares off casual players who might be looking for a challenge. A dual-core CPU is going to see more widepsread use than the PPU.

Just my 0,02 " ...

OldMan____
07-26-2005, 04:13 PM
Using a dual core processor and a separate thread is easier. Specially since the PPU would be used ina separate thread anyway!!! The PPU SDK proposed by ageia works on batch calls. In other words.. you feed lots of data.... then call some heavy weight operations that advance the simmulation etc... This must be done in multithreaded.

Multicores might not get same performance since a dedicated PPU may use large coherent memory caches (main memory is no coherent.. is maped in pages). And a PPU processor may be cheaper since it does not need complex brach predictions.

BUT PPUs face the problem of BUS.. PCI-E 1x and PCI are NOT ENOUGH for really heavy stuff. And not much people have an spare PCI-E 4x.


This fight is same fight of XBOX 360 and Playstation 3. XBOX 360 uses a multicore power PC (3 cores) and PS3 uses a single generic processor helped by a few sinergic processors (quite same kind of stuff that AGEIA PPU does).


BTW.. I am a game developer.. and here at my job we are not even thinking on using ageia solution anytime soon. We musty first wait and see who will be victorious in console battle.

FCIProject
07-27-2005, 07:50 AM
Interesting to here from a Games Developer.

That's an intersting point about the PCI bus. I had never thought of that.

It has been a longtime since I had to worry about hardware, like the bus, not being fast enough for the software I write, but you Games Developers must always be coming up against hardware shortcomings.

I am always totally amazed at the criticsm Oleg and his team face regarding IL2. I don't think most people realise how hard it is to write such complex software and the skill it takes.

IL2 has always been amazingly stable and has always pushed the Flight Simulation bounderies.

I was a great fan of Falcon 4, even though it crashed a lot, but IL2 is better than Falcon 4 in most respects now. Apart from not having a dynamic campaign or 'living war'. Install F4 and have a look if you don't believe me http://forums.ubi.com/groupee_common/emoticons/icon_wink.gif

This is why I believe hardware like the PPU can push games further. With multiple CPUs and dedicated processors like GPUs and PPUs so much more can be acheived.

OldMan____
07-27-2005, 12:33 PM
Just for sake of information.. hera at office me and some other developers put PF/FB in top 5 ranking of best games ever from developer point of view (considering complexity vs performance vs stability ).