• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Ageia PhysX & You

FrgMstr

FrgMstr

Just Plain Mean
Staff member
2FA
Joined
May 18, 1997
Messages
58,393
Ageia PhysX & You - Lost in arguing press releases and rhetoric spewed forth about Ageia's new PhysX PPU cards? We cut to the chase and tell you what you really need to know about PhysX and your money.

With the current titles available, PhysX delivers nothing worth the purchase price and PhysX has to be solely valued by the gaming content that supports it. Sites that are dragging the framerate wars into this are not telling the gamers what they need to know. Sadly this is exactly what I expected. PhysX is not about framerates, synthetic benchmarks, or PR catfights; PhysX is about the gaming immersion and experience it provides you.
Click to expand...

Read and dicuss. It looks to be a fun topic for sure. Let's be civil and keep this on topic please.
 
Definately not worth the price of entry at this point. Physics will play a big part in gaming in the future, but whether or not PhysX will be there is another matter. It will be interesting to see how well quad core CPUs can handle physics in software, as compared to a hardware solution.
 
The article isn't fair. Cellfactor is a better example of Ageia, GRAW just had Ageia support tacked on at the end, it's not 1/10 of what we should see with Ageia
 
Yep I agree with what you said in the conclusions. Right now it is pretty worthless, but until games come out that totally utilized and take advantage of it, then we can judge it.

To get me to buy one, I will see how UT2007 benefits from it.

PhysX does bring in many possiblities that developers can start to use in there games. Destuctable environments is probably one, which is something we all will love to have in a game like BF2 or something like that. Cellfactor does look it be fun for sometime, kinda like HL2 Deathmatch but will get boring fast. To me it looks like major tech demo.
 
Gob said:
The article isn't fair. Cellfactor is a better example of Ageia, GRAW just had Ageia support tacked on at the end, it's not 1/10 of what we should see with Ageia
Click to expand...

Regardless, an add-on card should both present improvement in both new and old methods of programming. Ageia has the resources available to accomodate the latter; if they fail to do so then their product is, ultimately, worthless, for it then is only beneficial in programs specifically designed for its use. As an analogy, would you expect your processor to only work in only one specific motherboard because the other motherboard manufacturers are busy accomodating the mass of people, those who use both old and new technology, rather than those who ONLY use that which is newer?

**Note, do not attempt to use this analogy to formulate a counterargument in which different processor sockets, through upgrades, are created, for they are not proclaimed to be an ultimate solution, rather an optional upgrade, but, unlike Ageia, supportive of older technology (ie agp/pcie/ide/sata). The analogy simply serves to put my former statement in layman's terms.**
 
I agree with much of the article, especially how Ageia should be credited with ushering in the big push to bring realistic physics into games. I don't really see the GPU companies posing quite as much of a threat to Ageia as AMD/Intel, although ATI/Nvidia are hyping their GPU physics a lot more. Both ATI and Nvidia are relying on the CPU to handle gameplay physics. Ageia could find itself between a rock(AMD/Intel handling gameplay physics), and a hard place(ATI/Nvidia handling the effects physics) Trying to face down ATI, Nvidia, AMD, and Intel on two different fronts simultaneously is indeed a tall order for any startup. Like the article says, it can be done if enough good games use and benefit from it.

The 3dfx comparisons are not so unfounded, especially since Ageia has brought in several former 3dfx guys.

I'm really interested in what game companies are investing in Ageia. I'm also curious to know why they are doing it. From the dev side of things, using Ageia's PhysX API is MUCH cheaper than using Havok physics, especially when the Havok FX license costs additional money on top of Havok's regular CPU physics.

However this plays out, its good to see the game world really phocus on physics.
 
or Havok could change their pricing scheme if such a standardized api became sucessful.
 
I think cellfactor is as good as it is going to get for a while (two years?). Developers wont make interesting ppu games without widespread adoption and there wont be widespread adoption with high prices and minimal content.

Short of a miracle hardware solution (cpu extensions / cheap physx cards) I foresee a slow adoption for massive physics in general. While ageia is fighting for widespread adoption, other physic touting titles will try to squeeze every ounce of performance they can muster in traditional ways. Either way ageia did a huge favor for gamers, I'm sure that in five years we will all take massive physics for granted.
 
Though I take slight issue with the framereates comments. Framerates *are* the point to gamers, in general, and particularly when using a product like PhysX takes bad framesrates (GRAW) to abysmal levels. And until there is evidence that this is, as Ageia and others claim, because of rendering more objects bogging the GPU (e.g., a pre-rendered with/without benchmark), I'm not buying that line.

Ageia has made marketing claims of how GPU physics won't work because the Ageia product has huge memory bandwidth, etc., etc., but present no evidence that this is in fact even needed. My porsche can do 193MPH, but my lotus will spank it on a track. There are too many factors involved for me to fall for the marketing BS, frankly, and until I see solid evidence otherwise, as opposed to marketing fluff, I too will be betting against the success of the company and product, while remaining a believer in the use of processing power to enchance the physics of games.

R
 
heycarnut said:
Though I take slight issue with the framereates comments. Framerates *are* the point to gamers, in general, and particularly when using a product like PhysX takes bad framesrates (GRAW) to abysmal levels. And until there is evidence that this is, as Ageia and others claim, because of rendering more objects bogging the GPU (e.g., a pre-rendered with/without benchmark), I'm not buying that line.
Click to expand...

As the article says the whole idea behind this is to improve immersion and the gameplay experience. If however to achieve this it lowers the framerate to unplayable levels where you have to decrease other "immersion" settings like resolution and or in-game graphic quality settings, then I agree this is something gamers aren't going to opt for and defeats the purpose of the whole "improve immersion" goal. In that respect framerate is important to look at.
 
I genuinenly have no pre-disposed feelings towards the PhysX boards. If it would be beneficial (from a game experience standpoint) then I would absolutely buy one.

From what I'm reading, sounds like we're not at that point yet.

Another point in the article (which is a really good one) is the idea of multi-core CPUs and how they ultimately help decide the outcome. I can't help but think that with quad-core CPUs coming (not to mention Quad SLI), that the PPU idea will end up being a novelity.

I guess we'll see.
 
pfc_m_drake said:
Another point in the article (which is a really good one) is the idea of multi-core CPUs and how they ultimately help decide the outcome. I can't help but think that with quad-core CPUs coming (not to mention Quad SLI), that the PPU idea will end up being a novelity.

I guess we'll see.
Click to expand...
I highly doubt quad SLI will have much of an impact on PPU sales. The number of people who will pursue quad SLI solutions will be very small. The cost of getting 4 video cards is much higher than most people are willing to spend.

Quad core is still also far away as dual core CPUs are still very new and when quad core is first released it will be very expensive.
 
Quad core will be here next year, and I doubt it will be any more expensive than dual-core CPUs are today.
 
Isn't ut2007 meant to be making use of these physics card? Mind you any vids ive seen of 2007 just shows the usual scenario of rockets slapping a wall and leaving the usual black mark and nothing more. :confused:

Edit: some tech demos on aegias site using unreal engine 3

http://www.ageia.com/physx_in_action/tech_demos.html
 
Gob said:
The article isn't fair. Cellfactor is a better example of Ageia, GRAW just had Ageia support tacked on at the end, it's not 1/10 of what we should see with Ageia
Click to expand...

Actually, it turns out that Havok was the one tacked on. The PhysX was the primary API used.
 
You will see PhysX hardware required levels in UT2007.
 
Physics will continue to become more complex in the future - with or without PhysX. So perhaps the PhysX is better suited to be a hardware physics accelerator for software physics (as opposed to adding physics effects - which ultimately bog down the rest of the system). This would offload physics off the CPU and lead to an increase in performance - kinda similar to how people use hardware-accelerated audio nowadays. However, since current game physics is very primitive, PhysX is way ahead of it's time - currently, physics probably doesn't take up too much CPU resources to make hardware acceleration worthwhile.

Of course, it's debatable whether hardware physics is even necessary with increases in CPU cores and speed. I guess this mostly depends on the suitability of general purpose processors for physics work. If dedicated and specialized hardware is needed to effectively process massive amounts of physics data, then maybe Ageia just pioneered something that's necessary in the long run anyway.

But will Ageia persevere? Only time will tell...
 
Sly said:
Actually, it turns out that Havok was the one tacked on. The PhysX was the primary API used.
Click to expand...

Where did that come from?

GRAW 360 had Havok tacked on for multiplayer ragdolls - it uses PhysX in singleplayer for everything.

GRAW PC had Havok for everything and tacked on PhysX for effects.
 
Capability aside for a moment.

Why are people so dead set against PhysX while saying Havok is the way to go?

A lot of people are saying that PhysX is a closed standard and is the next Glide, but that's hardly true - Ageia have hinted the PPU may accelerate other API's in the future.

Will Nvidia/ATi ever accelerate anything but HavokFX? Or will HavokFX be dumped in favour of microsofts far away standard?

We haven't even seen a game developer declare that they are using HavokFX yet. Let's look at the developers' point of view, in order to pay for something they have to be sure that it definitely adds a benefit for gamers to justify that cost. Everyone's already complaining that the extra effects from PhysX will cause slowdown and imply Havok will be immune and this is simply not the case. How long will it be before Ageia's competitors hit the market?

PhysX may well be with us in underwhelming force now, but the lag between here and when havokfx comes to bear may be enough time for PhysX to prove itself.

Edit: I love cellfactor. /pats teh BFG card. Seeing something first is worth every penny to me :)
 
MrNasty said:
Where did that come from?

GRAW 360 had Havok tacked on for multiplayer ragdolls - it uses PhysX in singleplayer for everything.

GRAW PC had Havok for everything and tacked on PhysX for effects.
Click to expand...

Kinda weird isn't it? The xbox360 version has the PhysX as primary, and yet chose the Havok for the PC version where it's supposed to be the one with the PPU.
 
Sly said:
Kinda weird isn't it? The xbox360 version has the PhysX as primary, and yet chose the Havok for the PC version where it's supposed to be the one with the PPU.
Click to expand...

Probably due to licensing fees.
 
Sly said:
Kinda weird isn't it? The xbox360 version has the PhysX as primary, and yet chose the Havok for the PC version where it's supposed to be the one with the PPU.
Click to expand...

It seems like a pretty stupid way of doing things doesn't it. :(

2 different developers - 2 different ideas of how to do things I guess :rolleyes:

I guess XB360 adopting PhysX helped a little too - Ageia has been working with microsoft to optimise PhysX for the 360's 3 CPUs and from how GRAW 360 plays they did a great job.

EDIT: eastvillager - Everyone knows that UT2k7 will be crap and no one will play it anyway. UE3 engine 4tl. /end sarcasm.
 
Fact is GRAW uses a tiny fraction of the PPU's power and yes it looks like it was bolted on at the end and it probably was!

This is the birth of a new technology and along with it comes all the teething problems that are always associated.

I agree with the article so far but It's all negatives and nothing positive (well not quite true but it's certainly not balanced)

Time and time again I see people fail to draw parallels to similar hardware, we don't all bitch and moan that the new £350-400 video cards are not worth the price because there's no shader 3.0 games when they're released and in fact 90% of the new features they have are useless untill something is created to show them off.

I believe their enterance into the market will start with added effects for games since these can be turned off with no impact to gameplay therefore dev teams can impliment these kind of extras without alienating any of their user base. Not everyone is willing to pay extra for better graphics but theres certainly a userbase who will, look at all the enthusiasts who have SLI, over twice the price of normal single card solutions for better graphics, you're telling me these people won't pay £200 for card which has the potential to add much nicer more realistic graphical effects.

Yes, the article is right on the nose, as always around here, but it's jumping on this and trying to kill it before it has chance to show off what it can really do, which is not helping the situation any. Yes users need to be aware of what they're buying into but they shouldn't be scared away from the thought of ever coming back.

Now the PPU's are out we can start looking for titles to use them, since the physics engine is free and has more features than Havoc it's more tempting for developers to use it, they get a free physics engine and the capability to add effects to their games for PPU users, this expands the games features and makes it more attractive to certain users, while being reasonably easy to impliment.

Which kinda brings me onto my last point, these physics effects are reasonably easy to code into the engine as I understand, its really just an extension of the normal physics engine. Games are becoming harder to make and take longer which costs more money, if we can use physics to add to game suddenly we have ways of doing things which is much easiser, we setup our "playground" of objects how we want them and let the physics run it for us. Think of the amount of time saved now a days with not having to animate characters deaths, simply make them a ragdol and BLAM almost totaly realistic movement with practically no effort. Apply this to lots of other scripted sequences or areas and you suddenly relieve a lot of work load.

It this kind of progress in technology which make games easier to make and thats always a big bonus for everyone, and shouldn't be ignored. For this reason I think physics (and AI) is going to become and increasing larger factor in the next gen games.

*edit*

A quick additional note here, I was under the impression the physics card can be used for running AI quite well, as a mapper I must say that I am looking forward to the day where I can simply put an enemy spawn point in a map and list the basic functions of the map (lifts, doors etc) and then simply let the AI learn the layout for itself, see where it can and cannot go and learn its environment rather than having to place path nodes everywhere which is incredibly tedious and really frustrating and lengthy to fix bugs in.

Another good example is creating random piles of mess, instead of hand placing hundreds of pieces of garbage or piles of rock and having to position and rotate each one to appear random I can just chuck in an emitter which spews this stuff out at the start of the map and it randomly falls about itself, saves me hours and hours of work.
 
Great article. Your impression is dead on.
I think its hard to justify the cost of the PPU right now, but I'm sure the same thing was said about TV when there were only 2 channels.

I see this physics model infultrating all aspects of games in the future, whether it be Havok or Aegia's idea of what physics is. Yes there is already a level of physical immersion in games, but this new model can only help to make things more interesting.

In my humble opinion the BFG Aegia card will probably die on the vine, due mostly in part to a lack of games, and lack of mainstream use. However the increased awareness will no doubt benefit games in the future. (that doesnt make alot of sense, but I hope you catch my drift.)
As always, solid article and no BS.
 
eastvillager said:
...and you won't see those levels on any populated public server for more than a day.
Click to expand...

More to come on this subject too....
 
If they are indeed multiplayer maps I would be interested how they are handling the physic objects. (assuming there are *that* many)
 
MrNasty said:
Will Nvidia/ATi ever accelerate anything but HavokFX? Or will HavokFX be dumped in favour of microsofts far away standard?
Click to expand...

R.E. the GPU, there is nothing stopping game developers from accelerating physics with another API besides Havok. It is just that NV struck up a relationship with Havok. ATI hasn't really done this, they are working directly with game developers to open up hardware access of their GPUs to developers to let developers choose how they want to accelerate physics on GPUs. Technically this is the case for both NV and ATI GPUs. So it doesn't HAVE to be Havok. It doesn't really have to be anything, it can be a proprietary physics engine as long as the game developer has access to hardware. DX10 will really push this forward on the GPU side.
 
Frosteh said:
Yes, the article is right on the nose, as always around here, but it's jumping on this and trying to kill it before it has chance to show off what it can really do, which is not helping the situation any. Yes users need to be aware of what they're buying into but they shouldn't be scared away from the thought of ever coming back.
Click to expand...

The article tells the situation as it currently is. Any negative connotation is due to the current status of games and the PPU. It isn't trying to kill anything, it is simply telling it like it is. It is what it is, and the article is very accurate on the state of things.

Frosteh said:
Which kinda brings me onto my last point, these physics effects are reasonably easy to code into the engine as I understand, its really just an extension of the normal physics engine. Games are becoming harder to make and take longer which costs more money, if we can use physics to add to game suddenly we have ways of doing things which is much easiser, we setup our "playground" of objects how we want them and let the physics run it for us. Think of the amount of time saved now a days with not having to animate characters deaths, simply make them a ragdol and BLAM almost totaly realistic movement with practically no effort. Apply this to lots of other scripted sequences or areas and you suddenly relieve a lot of work load.
Click to expand...

You have a good point here, will it increase game development cycles? I would say most certainly yes. But in the long run it can make for a much better game if physics is done right, the potential here is mind blowing.
 
Brent_Justice said:
Technically this is the case for both NV and ATI GPUs. So it doesn't HAVE to be Havok. It doesn't really have to be anything, it can be a proprietary physics engine as long as the game developer has access to hardware. DX10 will really push this forward on the GPU side.
Click to expand...

Meaning until DX10 if developers want effects physics in games they have 2 choices:

a) shell out for Havok

b) develop their own mini-api with help from nvid/ati

The second would be time consuming and, my guess is, in that respect more expensive than the first. Lack of options = proprietism as we see with Windows every day - having an unbalanced choice is the same as having no choice, especially where money is concerned.

And until Unified Shader Architecture is standard there won't be any gameplay physics happening on GPU's.

I think there's a lot of time for Ageia to perfect both the PPU and PhysX. Maybe they might make PhysX work on the GPU in addition to the PPU? Interesting thought.
 
Brent_Justice said:
You have a good point here, will it increase game development cycles? I would say most certainly yes. But in the long run it can make for a much better game if physics is done right, the potential here is mind blowing.
Click to expand...

I think he was saying it will make dev's lives easier through less scripting - helping to bring developement cycles back down in time...
 
Mark Rein had a didn't say much about the PhysX-required levels in UT2007 in a pre-E3 interview with IGN, but he did seem to hint that there definately will be some
We're also hoping to have some levels that really stretch the boundaries of physics. We have a good relationship with Aegea and what we feel is very good physics performance on PS3 and the possibility of putting the PhysX card to either match that or surpass that on PC. So we definitely are going to build some funky levels around are built around cool physics stuff. We haven't decided exactly what those are going to be quite yet so it'd be a bit premature to really comment on those yet.
Click to expand...
 
MrNasty said:
I think he was saying it will make dev's lives easier through less scripting - helping to bring developement cycles back down in time...
Click to expand...

I think when it comes down to it. Implementation wouldn't be any different from what being done now. It's hardware accelerated, but it's still the same PhysX API whether it's hardware or software mode. Development time wouldn't be any shorter whether you're using Havok or PhysX.
 
only one thing is nagging at me for this whole physx thing. i can't decide weather i should eventually buy either a physx card or another 7900GT to SLI. if games are going to start offering physics big time, then i want to be able to run it. i've gotta wait anyways but i have the feeling i'm going to have to make a call that i don't want to sooner or later.

why not get both you ask? well, here's the thing - i have the asus A8N32-SLI board and i have an x-fi in one of the pci slots. if i were to have 2 7900GTs i'd have to use stock noisey cooling on them both otherwise i'd have NO slots to spare for the physx. there goes my plans for sticking a nice zalman cooler on them. i was going to buy one last week, but i figure it might be smarter to wait it out and see what happens with physics support in the next 6 months. i want a crystal ball!!
 
This lot reminds me of back when the original nVidia GeForce was released, and it was criticised by many because it was often unable to deliver better framerates than the voodoo 5's, the Kyro II and the TNT2 ultra, in many games.

It was, of course, a pure fillrate comparison. History has demonstrated that nVidia's vision of add-on processing, was the way forward.

The PPU is certainly not worth the money now (just as a GeForce wasn't, back when it was released), but it will be.

The PPU is simply another step on the road to parallelism, and it deserves a place in the sun next the GPU and the multi-core CPU.
 
its definately interesting times for us gamers and hardware enthusiasts. i think the question on my mind and a lot of others is "will the PPU be worth my money in the next year or will havok and SLI take care of that well enough?"

NOW GIMMIE A CRYSTAL BALL!!!
 
One thing about this situation that gives me pause is the real vs. theoretical bandwidth issue. The official Ageia story for the slowdown in GRAW with PhysX is all the new data that the GPU has to display, but the Havok guys dispute that contention rather convincingly. Isn't it more likely, as has been speculated in these very forums, that latencies inherent in getting the data to and from the PhysX card to the GPU and CPU are creating massive bottlenecks?

Another side of this arising from some of the statements coming out of Ageia lately--they say that the internal structure of their PPU includes many independent processing cores and then they claim an astronomical number of calculations per second. Are they deriving this theoretical maximum by taking the calcs/sec of a single core and multiplying it by the number of cores? This would be standard marketing practice, but the practical reality would be very different. It's almost impossible that those cores would be used efficiently enough to get anywhere near the ideal max. It's even conceivable that there might be situations where only one core gets used, especially if the game developers don't granularize(?) their physics properly. All of this to say, what if PhysX is inherently a screamer on paper but a boat-anchor in real life?

We've already got examples playing themselves out in other processors. Dual SLI brings significant gains, but nowhere near 100% scaling of performance. So far, Quad SLI is showing sharply diminishing returns. We don't have quad-core CPUs yet, but the performance-scaling curve on dual-core CPUs is very similar to dual SLI. What if the "many cores" of the PPU face the same limitations, and only 2-4 of them are doing any useful work? Now, maybe the PPU cores function more like pipelines in a single GPU, which scale very well because of the parallelism of the work they do. But maybe not--anyone know what the parallelism is of physics calculations vs. display calculations?

Anyway, these are the warning signs that people's hands-on experiences with PhysX are raising in my thoughts.
 
Brent_Justice said:
The article tells the situation as it currently is. Any negative connotation is due to the current status of games and the PPU. It isn't trying to kill anything, it is simply telling it like it is. It is what it is, and the article is very accurate on the state of things.
Click to expand...

Yeah and with most of the negatives I agree, but it only seems to be telling half the story as if you never expect it to take off. I think the consumers deserve to see both sides, some foot work into finding out more positive stuff certainly couldn't have hurt.

You have a good point here, will it increase game development cycles? I would say most certainly yes. But in the long run it can make for a much better game if physics is done right, the potential here is mind blowing.
Click to expand...

You misunderstand here, I'm saying that I think it will decrease development cycles eventually, the engine only needs to be built once but the application of the physics within all the games based off that engine will benefit hugely from not having to script events.

Scripting events to seem random is very hard, especially when animation of free bodies is involved, not to mention that animating stuff takes ages anyhow. If you can setup a scene in aproximatly the right way and let the physics take over not only is it much easier to make and faster for the game developers but you also have the option of it happening in slightly different ways each time it's played which adds to replayability.

Development of game engines might very well get longer however.

*edit*

I think when it comes down to it. Implementation wouldn't be any different from what being done now. It's hardware accelerated, but it's still the same PhysX API whether it's hardware or software mode. Development time wouldn't be any shorter whether you're using Havok or PhysX.
Click to expand...

I was not really comparing Havok and PhysX directly, more like comparing advanced physics and what we had leading upto the release of PhysX.

Having the use of a much more powerful physics can speed up the time it takes to make a game in a fair few ways, we've already seen it help cut development times in certain areas of games already. I'm suggesting that having more powerful physics will again help cut down on development time in yet more areas.
 
You must log in or register to reply here.
Back
Top