Rumors: Ageia bought out

[AmberClad]

Usual Fudzilla disclaimer applies. Speculation is either Intel or Nvidia. Assuming it's true, I'd wager it was Nvidia, due to Intel's previous acquisition of Havok.

 

[Acanthus]

Nvidia has had interest in Ageia in the past, but they never came to an agreement.

I believe it was about licensing something so they could do HW physics acceleration on Geforce Graphics cards.

 

[lopri]

Could be Intel. They bought Havok last year so if it turns out to be Intel that bought Ageia, they can effectively lock NV/AMD out of lotsa patented IPs related to physics.

 

[CP5670]

That seems to have been a much better purchase given Havok's popularity. Most FPSs these days use some form of Havok while hardly anything uses PhysX.

 

[Sureshot324]

I doubt it was Intel. That would cause a lot of antitrust problems since they would have a monopoly on the gaming physics market.

 

[aka1nas]

Originally posted by: Sureshot324
I doubt it was Intel. That would cause a lot of antitrust problems since they would have a monopoly on the gaming physics market.

There is no gaming physics market yet.

I doubt it was Intel as they have little to no interest in a discrete physics accelerator and they already have a working software-driven API via the Havok acquisition.

 

[taltamir]

i doubt it was anyone. Investors don't like flushing money. And that is what aegia is right now. Fudzilla, enough said.

 

[nitromullet]

Originally posted by: taltamir
i doubt it was anyone. Investors don't like flushing money. And that is what aegia is right now. Fudzilla, enough said.

I dunno. Aegia is a gem in the rough...

The idea of a dedicated PPU intrigued a lot of people, which shows that there is a potential market for something like this. As I understand it though, the Aegia PPU essentially became a "Physic Decelerator" due to the fact that it was on a PCI card on the PCI bus. If you could get that same concept on a component that has a lot of bandwidth, say on chip (Intel) or on a device plugged into a PCIe 2.0 16x slot (NVIDIA) you might actually see it do something cool.

 

[n7]

Originally posted by: Sureshot324
...since they would have a monopoly on the nonexistent gaming physics market.

Repaired

 

[taltamir]

it intrigued me as well, but it took too long to arrive. There were absolutely no games that were using it. and in the end CPUs cought up, it makes more sense to pay for a better cpu then dump money into a dedicated PPU..

Now if they had a deal where you could SLI/CF two cards where one was doing graphics and one physics, and with them being able to just SLI graphics when physics was not avaialable... that would have really helped... that and actually getting games that work with it.

only recently did the first thing to ever need it come out... the tornado map in UT3.

I guess it could make sense of a company bought it, made a newer, faster physics card, and payed some developers to make games that use it. If there were games there will be a demand. But as things stand it makes no sense.

 

[theprodigalrebel]

I have been reading Fudzilla since its launch and I think they have been a pretty legit source (just like the other reputed tech sources). It's too bad Fuad can't shake off his Inquirer-legacy.

 

[toadeater]

I hope Intel does something with Havok rather than kill it off. I'd like to get some use out of my CPU instead of dumping everything on Nvidia's overpriced graphics cards with the broken jet-engine coolers and the crappy drivers.

Ageia was an ill-conceived design to begin with because it relied on the PCI bus. Not to mention it was too expensive. Nvidia could use SLI for physics, but how many people would buy SLI systems? If you're using SLI for physics, then you'd need 3 cards for SLI for graphics. You know that adds much more trouble and cost than just getting a quad-core CPU and running Havok on that.

 

[Sureshot324]

Originally posted by: n7

Originally posted by: Sureshot324
...since they would have a monopoly on the nonexistent gaming physics market.

Repaired

When I say gaming physics market, I mean software as well as hardware. Havok is charging money to developers to license their physics API, so therefore there is a gaming physics market.

Ageia's PhysX API can work in pure software as well as in hardware mode. This is why I think Ageia has a chance. They let developers use their API for free, unlike Havok. If their API is close to as good as havok's, many developers will use PhysX just because it's free. Once there are a lot of games out there using the PhysX API, the PhysX card will be a lot more attractive.

Even if Physics cards never become a big seller, I think Ageia can survive as a niche market. They won't need to invest much in R&D since they have no competition.

 

[taltamir]

Originally posted by: toadeater
I hope Intel does something with Havok rather than kill it off. I'd like to get some use out of my CPU instead of dumping everything on Nvidia's overpriced graphics cards with the broken jet-engine coolers and the crappy drivers.

Ageia was an ill-conceived design to begin with because it relied on the PCI bus. Not to mention it was too expensive. Nvidia could use SLI for physics, but how many people would buy SLI systems? If you're using SLI for physics, then you'd need 3 cards for SLI for graphics. You know that adds much more trouble and cost than just getting a quad-core CPU and running Havok on that.

Actually its even better... you buy 2 cards... in a physics intensive game 1 card calculates physics, one card calculates graphics. In a non physics intensive game OR in a game that does not support phyics, both cards calculate graphics.

This will actually add value to SLI systems AND make buying physics more attractive (since the second card would be doing grahpics when not physickying )

 

[ViRGE]

Originally posted by: taltamir

Originally posted by: toadeater
I hope Intel does something with Havok rather than kill it off. I'd like to get some use out of my CPU instead of dumping everything on Nvidia's overpriced graphics cards with the broken jet-engine coolers and the crappy drivers.

Ageia was an ill-conceived design to begin with because it relied on the PCI bus. Not to mention it was too expensive. Nvidia could use SLI for physics, but how many people would buy SLI systems? If you're using SLI for physics, then you'd need 3 cards for SLI for graphics. You know that adds much more trouble and cost than just getting a quad-core CPU and running Havok on that.

Actually its even better... you buy 2 cards... in a physics intensive game 1 card calculates physics, one card calculates graphics. In a non physics intensive game OR in a game that does not support phyics, both cards calculate graphics.

This will actually add value to SLI systems AND make buying physics more attractive (since the second card would be doing grahpics when not physickying )

It's all a pipe dream anyhow, no one has yet to use the GPU for physics in a game, and there aren't any games in the development pipeline with that functionality planned either.

 

[KlokWyze]

why does every fucking IT company with an original idea sell out?

 

[taltamir]

and 8 core cpus are just around the corner...

 

[Genx87]

Originally posted by: nitromullet

Originally posted by: taltamir
i doubt it was anyone. Investors don't like flushing money. And that is what aegia is right now. Fudzilla, enough said.

I dunno. Aegia is a gem in the rough...

The idea of a dedicated PPU intrigued a lot of people, which shows that there is a potential market for something like this. As I understand it though, the Aegia PPU essentially became a "Physic Decelerator" due to the fact that it was on a PCI card on the PCI bus. If you could get that same concept on a component that has a lot of bandwidth, say on chip (Intel) or on a device plugged into a PCIe 2.0 16x slot (NVIDIA) you might actually see it do something cool.

I think the best option is you will see the physics processor on the graphics cards pcb or integrated right on the gpu. PhysX was a technology before its time. I also think Microsoft needs to get involved and create a stansardized API before full fledged physics becomes the norm.

 

[hooflung]

City of Heroes/Villians, Unreal Tournament 3.0, Pirates of the Burning Sea and various PS3 games use Agea. There are others but those are the big ones. If Aegia was bought out as long as the new owner keeps it going it would be an awesome IP to own. If AMD bought it, well that would basically be the beginning of the End for Nvidia. AMD's name on it, despite their recent money problems, could put it over the top if it became a part of the Spider platform. Just think... Phenom's that could use 2 cores for physics as if there was a physical card... or a Crossfire setup that could use the 2nd 3870 to work as a physics card. Nvidia would have to build its own tech or liscense from its main competitors. Considering Nvidia is about to get hit hard in Northbridge market when Intel goes to memory controllers on their next generation they might just have to do something crazy like intel did when they adopted amd64 extensions.

That could all just be a load of crap, but I can dream in my fud fantasies.

 

[Munky]

IMO the whole dedicated ppu concept is a solution in search of a problem. For whatever physics a game is likely to use, a multi-core cpu is enough. We are not anywhere close to having games that model the real world with mathematically correct physics; that's just unnecessary and would consume enormous development resources that would be better served improving gameplay.

 

[taltamir]

Originally posted by: Genx87

Originally posted by: nitromullet

Originally posted by: taltamir
i doubt it was anyone. Investors don't like flushing money. And that is what aegia is right now. Fudzilla, enough said.

I dunno. Aegia is a gem in the rough...

The idea of a dedicated PPU intrigued a lot of people, which shows that there is a potential market for something like this. As I understand it though, the Aegia PPU essentially became a "Physic Decelerator" due to the fact that it was on a PCI card on the PCI bus. If you could get that same concept on a component that has a lot of bandwidth, say on chip (Intel) or on a device plugged into a PCIe 2.0 16x slot (NVIDIA) you might actually see it do something cool.

I think the best option is you will see the physics processor on the graphics cards pcb or integrated right on the gpu. PhysX was a technology before its time. I also think Microsoft needs to get involved and create a stansardized API before full fledged physics becomes the norm.

DirectX 11 has already been stated to define physics acceleration by video cards.

 

[BlueAcolyte]

AGEIA!!! IT'S AGEIA!!! NOT AEGIA.

end rant. I don't really see the point of a PPU anyway... The CPU is strong enough to calculate all the physics, but most importantly, the support is lacking.

 

[aussiestilgar]

Yeah I'd think that the move to quad core CPUs as the norm would see the advent of physics, AI etc. computation spread on other CPU cores rather than to the GPU. But whomever bought Ageia might be buying the API for themselves... Maybe even add it as a chip on the motherboard, that'd be cool...

 

[Genx87]

Originally posted by: munky
IMO the whole dedicated ppu concept is a solution in search of a problem. For whatever physics a game is likely to use, a multi-core cpu is enough. We are not anywhere close to having games that model the real world with mathematically correct physics; that's just unnecessary and would consume enormous development resources that would be better served improving gameplay.

We dont have it because we dont have a capable platform to do it. The PhysX card was the right idea with a bad implementation. However slap that PPU on a graphics card or on the GPU itself, elimintating the bottleneck and we are talking. I dont think Multi core CPU's are the answer either. They are more a bandaid than a solution. Much like MMX and SSE were a bandaid for CPU rendered graphics before dedicated graphics chips made their way into PC's.

 

[Sureshot324]

Originally posted by: munky
IMO the whole dedicated ppu concept is a solution in search of a problem. For whatever physics a game is likely to use, a multi-core cpu is enough. We are not anywhere close to having games that model the real world with mathematically correct physics; that's just unnecessary and would consume enormous development resources that would be better served improving gameplay.

Game developers develop games designed for the hardware available at the time. That is why game's demands of CPUs and GPUs stay balanced with the hardware that is available, instead of some games being massively GPU limited or CPU limited. If physics hardware became more mainstream, game developers would make games take advantage of it.

One area in the future that I believe will have a massive demand for physics calculations, is physics driven character animation. Ragdoll physics will apply to each character all the time, not just when they die, and there will be an AI controlling or helping to control the character's body, so there will be no fixed animations. There is already at least one company developing such a system.