Ageia's PhysX Needs a Killer App @ [H] Enthusiast

I think the valve hardware survey tells us how much they are really going to sell...... :rolleyes: . I mean they even have the enthusiast line of people such as us split up ina decision weather it is worth it or not! The usuall consumer probably spends about $500-600 on the computer itself, the tower only. They sure as hell are not about to spend $250 on a card to do physics when they are running onboard video or $100 video card for that matter.

I think it is the wrong thing to sell the physics card seperatly. I have seen people bash ATI/NVIDIA for their solution using a video card to do the work.....well this solution will help out in 95% of the situations as opposed to 1 out of 10 games that would benefit using the ageia card. I figure that if the video card on sli or crossfire is not being used for physics, then it will be able to be used for graphics. So this would be a win win situation. One card either helps the FPS and visual quality or it can do the physics work.

Antec 500 SP $50 after MIR
CPU 146 running at 3.0ghz $160
7900GT $300
DFI Ultra-D $110
G.Skill 2x1gb ddr500 $150

Ageia PPU **********$250*********Kiss my ass, not going to happen

So there, my $0.02 from somebody not about to pay $250 for this Ageia card.
 
mashie said:
And which game should they mod that already support the PPU?

We do have to wait for a developer to add the PPU as an engine option to speed-up a game somewhat. But we don't have to wait for those developers to require a PPU as a minimum to play their game. I figure they'll soon recognize PPUs and use them (inefficiently at first) for better first-person frame rates in an Unreal engine, for example. That is when the modding community can run with it for making multiplayer a different environment.
 
bildad, you're right, but what I'm trying to say is that it is up to the game developers to make games that make physics processing power matter. GPUs can use spare cycles to do physics work on a smaller scale, like what is seen in HL2. Ageia's PhysX is capable of so much more, but unless the developers figure out a way to make it matter on a larger scale in gameplay, market traction will be difficult. A "Killer App" is needed to get them started.
 
rjblanke said:
bildad, you're right, but what I'm trying to say is that it is up to the game developers to make games that make physics processing power matter. GPUs can use spare cycles to do physics work on a smaller scale, like what is seen in HL2.


The GPU as PPU solution gives you NO hardware accelerated gameplay physics...
Unless you count CPU calcualted physics as "PPU" accelerated...*chough*

Terracide
 
I didn't bother reading the thread, so this may have been covered already..

In regards to multiplayer, exactly how much transmission bandwidth will be required for all of those objects? It seems that such info could choke a walrus!

Even if they could allow each of the clients' PhysX cards to calculate the results using simple "this object strikes that object at x speed", imagine joining a game in progress! I would think game servers would need to revert back to the Doom1 days, where nobody may join a game except at the start of a match and if anyone drops out, the game is over. Either that, or the entire server forces a pause while new clients receive all of the current object physics info.
 
I've said it before, and I'll say it again...

The first vid card mfgr that puts a sticker on the outside of their box that says " Ageia PhysX PPU On Board!!111" is going to pwn all other vid cards.

Also, for you nea-sayers out there about the adoption of the card, all that has to happen is Futuremark puts a PPU score factored into the 3DMarks that raises the 3dmark score by a few thousand, then every 3dmark fan or junkie is going to HAVE to buy one. Just like when they make a new release for new shader modles and to even score above 3000 you have to buy a new video card.

I have a couple questions:
1) Can the PPU be used in other applications that require math, such as Folding@Home or SETI@Home, etc?

2) From what I unserstand, Havok only supports visual effects and in-world interaction of objects is non existant. The Ageia PPU makes everything interactive. Am I right? So why would you be content with "fake" physics eyecandy when you can have realtime object interaction?

3) Do you think that as supply goes up, price will fall? Early adopters ALWAYS pay a primium for having the latest gadget... but in 6 months, or a year down they road, when all (good) games being developed use the PhysX API, do you think the price per card will go down? I'm not opposed to going out and paying $250 for one, but I bet more people would pay $100-$150 for one.
 
glutto said:
I didn't bother reading the thread, so this may have been covered already..
Which is why n00bs should read the thread...

Bandwidth and cost aside, level developers have to show me a reason to buy another card other than the ability to throw boxes around. Ageia has shown examples of possibilities of what it could do, but in the video Kyle touted, I'm not being shown any true creative uses of the product other than lots of little things being thrown around.

It's like being shown gravity for the first time and all anyone seems to be doing is throwing apples at Isaac Newton in a deathmatch. There is much more that could be done.
 
You want the KILLER apps?
1. FOLDING AT HOME. if it helped some it would sell instantly.
2. UT '07 it should be a requirement from the screen shots.
Why can't AMDand Intell add a PPU to the CPU die? They have ALUs and FPUs.
What cant the PPU be on the motherboard like some sound chips, Most would pay the premium to not need to stuff another PCI card in.
 
The smart thing to do would be start using it for multiplayer apps right now where the card isnt needed on the client end. That way everybody could get exposed to it and see the benifits immediately.
 
heh... PhysX [H]orde Edition huh?

if the apps are there.... that is what i hear, and i would have to agree. there is no single app or game that would convince me to get one.

it begs the question, how many would it take to get you interested?

3-5? or more than 10?

i would rather get a X2 with the $250... but that's me.
 
In regards to everyone complaining about bandwidth. The whole point of the PhysX card is that the server DOES NOT have to send the movement data on all the objects to the clients. Jesus Christ that is what the card is for! The only reason no one can get past this is b/c there hasn't been a game yet, released or announced, with an elegant solution as to transferring basic information, then handing it off to guess what, the PhysX card that resides in your computer. If you think the server has to calculate all that, then you wouldn't even need the card in your computer, as the server just did all the work. Of course this would completely clog bandwidth. All we are waiting for is someone to realize that you simply transmit the initial (and heres the key) *simple* object interactions, and let the client handle it after that. This would require every computer that is playing have a PhysX card (or emulator) including the server. All the calculations are done internal to each computer, and the final results MUST BE THE SAME. That's just the way math works. It's objective, see? =)

Now we come to the 'trusted computing' portion. Yes I realize this gives hackers a much easier chance to ruin and raze, but if you think that is a serious reason to halt all forward motion in this area of gaming, plz awp yourself in the head ftw. I sincerely hope some devs pick up on this. Die hackers die, you ruin all my gaming fun. Peace im out.
 
kraemer007 said:
The smart thing to do would be start using it for multiplayer apps right now where the card isnt needed on the client end. That way everybody could get exposed to it and see the benifits immediately.
That's the problem, we'd need the card to see the effects. Also if you look at this clip of normal vs Ageia the damage model isn't the same.
http://physx.ageia.com/footage.html
 
Unfortunately it's the graphics upgrades that are going to sell the PhysX cards to gamers. It's been mentioned here that having gameplay-related features will break gaming communities, and I think that's absolutely true. This is why you can't relate the PPU to the advent of 3D cards. When I bought Half Life, I initially played it in software mode. It was one of the games that made me want a 3D card. But I could tell my friends to get the game even if they didn't have a 3D card and it was the same gameplay experience, and they could purchase a 3D card later to see some visual improvements. Slowly but surely they have become ubiquitous in computers, to the point now where you can say "3D Acceleration Required" on the box and have it be a non-issue (this would have seriously hurt games like Half-Life at the time). The PPU can't use this approach with gameplay features since you'd be marketing a different kind of game to users with and without a card.

What you will start seeing are things like a version of the Sims with realistic clothing animated by the PPU, or earth spells in a Warcraft game that collide thousands of chunks of ground together. You'll see more realistic flow out of a waterfall in the background of an RPG, or (as in the example linked above) much cooler vehicle explosions in an FPS. This eye candy will get the cards out there, and then you'll start seeing the games where the gameplay starts being affected. I see that as the only way that developers will become comfortable programming gameplay with the PPU: when they can say "PPU Required" and know the market share is there.

This is also the issue that I have with the video card companies' approach (this "effect-based physics"). If all they're worried about is the eye candy at this stage, they're missing the whole concept of what PhysX is trying to accomplish. People won't buy the PPU because they won't see what it offers over a GPU-based solution. So the video card companies are basically screwing us out of future improvements to gameplay.
 
Don't know if this was mentioned in this thread, but why the glowing preview of Nvidia's Physics implementation at http://enthusiast.hardocp.com/article.html?art=MTAwNSwxLCxoZW50aHVzaWFzdA== ??

Why no complaint about a killer app for Nvidia's method? Because Ageia wants you to buy a $250 card? Well, Nvidia wants you to buy another $500+ card for their physics.

Personally, I think Ageia is following the right path with their card. They have the developer support. They waited on the hardware launch until the games were up to snuff.

I also thought this was the enthusiast site. Why complain about new hardware when that is what the site is about? I mean we should all be excited over this new technology. I know I am. I can't tell you how long I have waited for the day where there are trully interactive and destructible environments in games.

*edit - punctuation*
 
Low roller has a link that confirms our fears with cell factor's multiplayer:

Low Roller said:
Cell Factor is a LAN-only game because of bandwidth issues over the internet.

http://www.firingsquad.com/features/gdc_2006_report2/

If anyone of you have tried to play a game that relies strictly on allowing the clients to do a load of the processing themselves you will find that everyone will get out of sync with each other in a relatively short amount of time even on a lan. You have to send coordinates and other client data (such as who the server thinks was hit or not) to all the clients so you end up with something that looks the same on everyone's screen.

The physics objects will need to have their reference frames sent to all the clients at some point. (In cell factor, where objects interact in a critical way, the server does need a ppu too) They could try various ways to limit the amount of data sent, but that will eventually lead to a lot of out of sync moments. (Like dropping dead when you nearly missed a crate or worse yet some one coming to life after you've already seen them die for half a second.) Finding an acceptable in between will be difficult, but not impossible.

(Knowing the server's final orientation of a crate a mile away isnt as important as the one you just thought killed the enemy aiming at you.)
 
mashie said:
Actually if all systems are PPU equipped then you don't need a single byte more traffic going to/from the server than it is doing with games today.
In regards to joining a game in progress, I can see this if the additional objects cannot interact with the players and are merely eye-candy, such as fancy flamethrowers, plasma streams, etc, but when each or most of these objects can damage and/or block the paths of players, I would think the precise info of these object for each client would matter a great deal.

By the time a new client received the info of the thousands of objects in a level, they probably have largely changed already, requiring their gamestate to be drastically updated again.

There's also the problem of lag. I would think that even a 100ms delay would be enough to skew the precise results required.

Good thing this isn't my problem to deal with. I just pity the poor dial-up user.

My comments are not meant as complaints. I'm a bit disappointed that some people interpret them that way, and also possibly interpret a simple post count to intelligence.
 
I can't wait to see people overclock their PPU cards, LOL!!

But seriously, there are a lot of things that have to happen for this thing to pan out like others have said. I have not read thru the whole thread, but hopefully the PPU will be able to be used among API's with-out a whole lot of trouble and with-out too great a loss in performance from one format to another. Will we see an implementation requirement in DX10b maybe?
 
I hope they make it , but the gut tells me that both ATI and Nvidia are going to come out with their own versions built in to their own video cards,or PCI independent for that matter and Ageia is going to become a Voodoo on the timeline.
Either way as a [H]ard core gamer it's as important as a X-Fi in a gaming machine because of what it will add to the great experience that is Gaming!
 
The head of the nail has been hit with this one! Put up or shut up... period! The folks at Ageia have the opportunity to be absolute pioneers! Not since the introduction of the graphics accellerator has there been this level of opportunity to make a deep-rooted imact on the IBM compatable PC platform.

Don't squander it Ageia! I for one will make the investment if you give me the hardware AND the killer apps.

Kudos to the [H] for pushing the buttons. It's about time we found out what PPU's are all about from a consumer perspective!
 
prtzlboy said:
Um, you're missing the whole point of the argument. My point is that the type of physics acceleration that doesn't REQUIRE a physics card is just more eye candy, and for me not worth the extra dough. More flying crates with no impact on gameplay. I want physics that affects the way the game is PLAYED not the way it looks, and this is the type of thing that you would need the hardware for, which is why it is a catch-22. It would be hard for someone to justify developing a game that would REQUIRE a $250 add-in card.

You in left field. Your confusing reality with what Gabe said. Re-read what I said, it makes Gabes point moot, these cards accellerate physics that DO affect gameplay.

Top that off with regular video card manufacturers incorporating support for this kind of physics in the video cards GPU. Obviously, it will slow the video card down, but it will work. This means ANY game can be made to use it as long as you have a card from ATi or nvidia, but you will get max graphics performance if you have a dedicated card for physics. No one will be left out in the cold.

This leaves Ageia as an engine licensing company NOT a hardware company. In short, the video manufacturers are helping Ageia.

Why don't you all get this? It's common sense business moves.
 
That looked far too much like a tech demo to me. It looked like they just put 10,000 boxes on the map and let everyone go on a spam fest with them. With individual player's God like control over his environment I can't see how this game could be anything but a float in the air spam fest.

I'd really like to see more creative implementations of this. Destructible vehicles and environments would be big here. After this comes out I never want to see a health bar on a vehicle again.
 
BBA said:
You in left field. Your confusing reality with what Gabe said. Re-read what I said, it makes Gabes point moot, these cards accellerate physics that DO affect gameplay.

Am I reading this right?
Are you trying t say that GPU's will offer gameplay-physics acelleration, and not just eyecandy-"physics"?

Terra - Collision dectecion any everything?
 
glutto said:
In regards to joining a game in progress, I can see this if the additional objects cannot interact with the players and are merely eye-candy, such as fancy flamethrowers, plasma streams, etc, but when each or most of these objects can damage and/or block the paths of players, I would think the precise info of these object for each client would matter a great deal.

Initially speaking, most of the objects will be eye-candy. Neat explosion effects, objects deforming certain ways, et al. Later on, in more sophisticated implementations, you may see it becoming more, but right now, no. It'll be kind of like the water in glQuake. Yeah it's neat, and maybe even unfair if you couldn't afford a voodoo, but, it's still just eye-candy.

By the time a new client received the info of the thousands of objects in a level, they probably have largely changed already, requiring their gamestate to be drastically updated again.

See above, it'll be unnecessary to even send it to unequipped clients (who can't do the math themselves).

There's also the problem of lag. I would think that even a 100ms delay would be enough to skew the precise results required.

The mini-gl version of quake didn't screw up the game too much with eye-candy, this won't either.

Good thing this isn't my problem to deal with. I just pity the poor dial-up user.

Why? All the math is done locally. If it's an unequipped unit, it'll never even have to care.

This is entirely client processed. If you don't have the card, you'll never see the effects nor have to process them. EOS.

My comments are not meant as complaints. I'm a bit disappointed that some people interpret them that way, and also possibly interpret a simple post count to intelligence.

Okay, umm, have a nice day? :D
 
Heh, been thinking of how to respond to this thread all day since I read it this morning.. think I got it...

GPU does all the graphics, everything you see on the screen, we all know that. Plain and simple if you're green or red, you're pretty damn happy with what they can do right now. That's fine and dandy, release faster cards with more pipelines with more memmory, etc etc. Maybe add a third player in there too (Hopefully, Would be SO much better as a three legged race, but thats another topic.)

PPU will do all the calculations for all the collision detection and environmental interaction. What that means is if it's foggy, you should disturb the air/fog as you move through it. GPU physics can't do this w/ Havoc. Liquids, moving objects, and atmospheric effects are going to be the name of the game for this. (No pun intended)

CPU; Heres where I'm getting pretty excited. We have Dual Core now and Quad core on the horizon. What if the AI for each of enemies you faced on screen had a CPU core dedicated to AI? I'm not talking about 3 or 4 decisions it makes depending on what you're doing. I'm talking full blown LISP API's doing the thinking here, with the ability to Shoot, call back up, Get pissed off when they run out of ammo and chuck their gun at you, Sidekicks that don't walk into your line of fire to take a bullet in the back of the cranium. ETC

Honestly, Ageia must pull this off perfectly, and I don't think there's going to be one killer app for this thing to succeed, there needs to be FIVE OR SIX that cover a multitude of gaming genre's. Not everyone likes FPS's, and I can see RTS's and RPG's actually benefitting the most from this. Why? Slower RTS and RPG game speeds donit need to send as much data over the internet as a fast paced action FPS game will.

That's a double edged sword, due to to the FPS genre being the most widely played in the U.S. And Honestly, I don't think ATI or Nvidia will kill this product. Lack of enough bandwidth between users will. That or the LAN scene is about to explode again, (and we all know we need to get out of the house every once in a while!!!)

On a side note, the PPU will eliminate the Client-Server networking model, and move it to a peer-to-peer model. Why? It would lower latency between players and fast moving objects, and be easier to keep in sync. I just don't think PPU enhanced FPS games can survive with a client-server design.

Right? Wrong? Time will find out.

And yea, I'd pay $150 MAX for a PPU. I'm not only a gamer, my wife is too.
 
drizzt81 said:
Which is why they need to Port F@H to the PhysX card!




Depends, there are two options:
  • server side physics, in which case each 'fragment' of an object becomes and object and its velocity and position is transmitted to all clients. This requires a LOT of bandwidth.
  • client side physics: the server sends the velocities, positions of all objects to the client (this is what I image CS:S does anyway). The client then computes any/ all physics interactions between the objects. Requires that the server can trust the client will perform this computation correctly (i.e. hacking may be easier). Let me elaborate: the clients would keep track of other object/ object interaction (i.e. player with fragment) and not the server, which is only interested in 'interactive' objects, such as bullets and players.

Half life 2 can do both and does do both.

HL2 DeathMatch very much relies on sycned physics to allow the gravity gun to be used as a weapon, but also any objects too small to be able to do damage (or block movement) are simply calculated client side.

Counter strike source uses this a lot, most of the tin cans and rubbish laying around in the streets is all using client side physics, and can be turned off or have its fade distance changed. However the larger objects such as barrels which player can collide with all run really badly, theres a really soupy movement feeling when aproaching a barrel as you slowly decelerate to a stop and then get pushed away, its not done very well at all, it certainly doesn't have the bandwidth overhead that HL2DM has.
 
If the next generation of UT uses it, I'll be buying one... Actually 3.

Hey Kyle ask John Carmack what he thinks of the idea....
It would be interesting to know.
 
lol i said exactly this in another forum, needs a killer game, thought maybe UT2007 would deliver, but there just isnt a game company out there ballsy enough to put out a revolutionary game, perhalps requiring a physics engine, and i do mean requiring, with that engine you would totally change gameplay, and it wasnt the video of those people flying around throwing crates at each other that impressed me the most, it was the short clip on their site of physx vs non physx.

we need a doom3 for physics, only doom3 was in a much better situation because the 3d acelerator was already pretty damn mature, i only say doom 3 cuz every card out at the time really could not run it with high quality.

but like iv said before, i think that there needs to be some sort of integrated physics engine, and if this does take off, where are the three slot PCI-e mobos! we are going to need em i mean eventually pcis 2gbps or w/e it is isnt gonna be fast enough. bring on integrated physics with Nforce 5! bring on integrated physics with Xpress 3900!

Frosteh said:
Counter strike source uses this a lot, most of the tin cans and rubbish laying around in the streets is all using client side physics, and can be turned off or have its fade distance changed. However the larger objects such as barrels which player can collide with all run really badly, theres a really soupy movement feeling when aproaching a barrel as you slowly decelerate to a stop and then get pushed away, its not done very well at all, it certainly doesn't have the bandwidth overhead that HL2DM has.

agreed. love source, not so much for the physics, if counter strike source.... source... (HL3) ever comes out i hope it has physics up the wazoo, but then steam doesnt seem to be letting go of the source engine for a while, and personally i think that the way the explosions are done is fantastic. not in the simple char stained surface, but in the flying of stuff around, already if you throw a grenade into a pile of cans, there are some pretty impresive results, also the bodies are great.

on the other hand, the ability to cause a beam to come crashing down, roofs falling over or stuff cathing FIRE (something not yet advertised on physx site) or the abillity to trip, would be REALLY cool. how would you like to be out side of some crazy cabbin with no way in for fear of head shot, so you just shoot the metal bracers and set the building ablaze. now there the ones who want out
 
BBA said:
This leaves Ageia as an engine licensing company NOT a hardware company. In short, the video manufacturers are helping Ageia.

Why don't you all get this? It's common sense business moves.
Perhaps because, as stated more than once in this thread, Ageia is GIVING AWAY the engine in order to make money on the hardware? In order for your scenario to work, they'd have to flip their whole business model upside down. Not to say they can't, but if we're talking about what they're doing right now, they can't be a licensing company, because they are giving the licenses away to prime the pump for hardware sales. If they try to change their model midstream, their partners will probably start dropping them like a bad habit.
 
Completely underwhelmed by the cell factor video. It was just to chaotic and pointless looking. And the Ghost Recon video is a sham. The PhysX explosion looks exactly like it does on the 360...and I imagine that's how it'll look on PC as well.

I've yet to see anything cool. They need to start showing some liquid physics possibilities or something.
 
Lot's of things are "possible"

What we need to see though is this stuff used in games to make us want it.
 
I for one am easily amused and find the demos appealing. I will purhcase a card as soon as it is possilble. I will also be running sli and in the event there are different games using different physics hardware /software approaches , I assume that along with a more powerful cpu, that i will be able to enjoy whatever games are devloped using both "technologies" if you will. . Ageia has sparked my imagination and I will most certainly invest in them.
 
Low Roller said:
Artificial Studios just today released another video for CF. This one shows a little more than the others.

CellFactor Bink Video - 398mb



(thanks SlamDunk)

Thats impressive. Assuming the game has more than one map and works fine over the net in all its glory I'm willing to buy the card for it. (Only problem is I'm broke after buying a fw900 today.)
 
I love my Verizon FIOS.

FIOS.jpg
 
Back
Top