What are phys x accelerators ?

kinggost · 8 Apr 2006 at 18:55

shame i cant fit these in between my sli watercooled cards or id have one

geeza · 8 Apr 2006 at 19:20

kinggost said:
shame i cant fit these in between my sli watercooled cards or id have one

can you buy pci adapters?

kinggost · 8 Apr 2006 at 20:23

geeza said:
can you buy pci adapters?

dont kno dont think so

Kamakazie! · 8 Apr 2006 at 23:27

Lanz said:
I dont think that would work niether, like someone said. You have the enhanced physics, but each one of those boxes (or whatever) has to be kept track of by the server, this would require vast amounts of bandwidth both server and client side compared to todays online games. And wouldn't the dedicated server also have to have one of these cards to process all this physics at the same speed as the client? So i think this technology is mainly for single player games, and would only increase fps for online games.

i dont see why the data for each box would have to be sent to the server/other player. Surely the world can be set and then only players mvoements/bulletts/actions need be kept track of because on each computer, these actions will interact with the environment in exactly the same way.
If player A fires a bullet at a stack of crates, only the bulletts location and movement need be sent as on each computer in game, this bullett will hit the stack of crates and do exactly the same damage...

the cell factor multiplayer video doesn't seem to have any problem with this, although obviously they could just show the video and ignore any problem if it arose.

bliss007 · 8 Apr 2006 at 23:29

Hmm, I wouldnt want to go back to my cards of lots of pci card, I like big empty space for lots of air, i will get one for Vista when they add the chips to Nvidia cards so its not 2 cards to do the job.

gord · 9 Apr 2006 at 11:02

Im interested in seeing the improvements people notice from their £200~ purchase.

NickK · 9 Apr 2006 at 11:23

lowrider007 said:
So you think that the ppu *may* even decrease you fps, tbh I, like proberbly many other people was thinking that using this ppu with a physX supported game you would get an increase in fps but that inceasingly looks like its not the case, by your reckoning then, the most we can expect is the same fps but with the added effects ?

Remember this is a physics card that does the calculations between objects.

Increase the number of objects and you increase the power required to render those objects - both in graphics cards and the main system resources. So the physics engine can calculate the next position for 1 million gas particles - the computer then needs to transfer those million particles to a graphics card for rendering.

It will be interesting to see how the games designers cope with a non-level playing field. It's bad enough DICE not implementing widescreen for BF2 under the exuses that it gives the player an advantage.
If there's more accurate smoke - it will move and look different if there's someone sat in it compared to low-physics model as it is now.

I'm hoping for the API to expose the card for non-gaming applications.. it would be nice if the API handled scaled number of cards too (assuming the BUS isn't going to saturate).

AtomicBanana · 9 Apr 2006 at 23:57

At that price? no ******* way :]

JimJones · 10 Apr 2006 at 10:05

I've been folowing these for months now and think that they're a great idea. AMD and Intle aren't going to design a core spercifcaly for physics. It's to restrictive for the CPU.

I would think that they will have PPU only severs or some other way of sorting it out. I wouldn't worry about it they will have thought it out.

Jim

HEADRAT · 10 Apr 2006 at 10:22

gord said:
Im interested in seeing the improvements people notice from their £200~ purchase.

Somewhere between zero and none ATM

HEADRAT

spookywillow · 10 Apr 2006 at 11:35

Lanz said:
I dont think that would work niether, like someone said. You have the enhanced physics, but each one of those boxes (or whatever) has to be kept track of by the server, this would require vast amounts of bandwidth both server and client side compared to todays online games. And wouldn't the dedicated server also have to have one of these cards to process all this physics at the same speed as the client? So i think this technology is mainly for single player games, and would only increase fps for online games.

this is the exact reason BF2 cut back on the destructable buildings, it was possible to make them but the bandwidth it would take to keep track of all the buildings parts etc would be imense as each client would have to know where each part of building landed etc as it would have to be identicle on each client.
________
vapor genie vaporizer

nands · 10 Apr 2006 at 11:35

I like the side by side video on the main website showing the game with and without. I would have to see a good deal more of those comparisons before coming even close to making a decision!

n3x · 10 Apr 2006 at 11:52

http://pcweb.mycom.co.jp/articles/2005/09/13/cedec/images/027.wmv

http://www.gdhardware.com/files/ageia/BuildingDestruction/side_by_side.wmv

Muzy · 10 Apr 2006 at 12:27

where is the OEM option gone ?

I can't find it....

n3x · 10 Apr 2006 at 12:29

Muzy said:
where is the OEM option gone ?

I can't find it....

Maybe sold out? LOL

macachia · 10 Apr 2006 at 13:21

oem version gone

i was going to order one today

puppy · 10 Apr 2006 at 13:52

Well

As mentioned elsewhere this could really enhance the realism of single-player games, and when they become reasonably priced (and hopefully passively cooled - not *another* fan :rolleyes:

) I'll certainly consider getting one if it's supported by lots of games that I want.

However, everyone knows that, right now, multiplayer gaming is where its at - and these new chips are going to be useless in that environment. Because of the amount of data that would have to be shunted between the client and server describing the trajectory, speed etc of all those discreet objects the bandwidth requirements would be astronomical and, as another poster has pointed out, both the server and *all* the other clients would need to have the chip on their systems for this to be functional anyway...

That being the case, I wonder exactly how good the takeup of this new technology is going to be...

Dave · 10 Apr 2006 at 17:03

Muzy said:
where is the OEM option gone ?

I can't find it....

Sorry to be a spoilsport, but 400x75 is the max image size in a sig. (yours is too big) Plus I'm fairly sure you aren't allowed animated ones

dEl_fUEGo · 10 Apr 2006 at 17:41

puppy said:
As mentioned elsewhere this could really enhance the realism of single-player games, and when they become reasonably priced (and hopefully passively cooled - not *another* fan ) I'll certainly consider getting one if it's supported by lots of games that I want.

However, everyone knows that, right now, multiplayer gaming is where its at - and these new chips are going to be useless in that environment. Because of the amount of data that would have to be shunted between the client and server describing the trajectory, speed etc of all those discreet objects the bandwidth requirements would be astronomical and, as another poster has pointed out, both the server and *all* the other clients would need to have the chip on their systems for this to be functional anyway...

That being the case, I wonder exactly how good the takeup of this new technology is going to be...

I see a few people saying that this ppu will only be useful in single-player games. UT2007 and GRAW both seem to me to be heavily multiplayer orientated, even the single-player side of GRAW can be played co-operatively, how would this be the case if bandwidth requirements will be too great when using the ppu? Will these games only be playable to peeps with SDSL lines of a minimum of 2megs or so? I have my doubts. I have no real understanding of netcode in game, but, as has been pointed out 'multiplayer gaming is where its at' and i trust that Agiea and the developers of such games that will use this new tech have thought about all the imposing problems that go along with it a long time before the games actual development began. If anyone can provide hard factual evidence that these games will be problematic in multiplayer sessions i would very much like to see it.

EDIT: I find it hard to imagine that we mere forum goers could come up with 'multiplayer problems' for this tech off the top of our heads that somehow the developers themselves 'might' of missed. Highly unlikely really.

Goksly · 10 Apr 2006 at 17:48

I dont understand why heavy use of phsyics will have any huge effect on bandwidth? If this was the case, then only one PC would need the phsyics card and then that would tell all the client machines where all the shrapnel has gone.

Its my understanding that its up to the clients to render the frame, so the bandwidth should still only carry the usual data -> this player is here, there has been a bullet fired here etc etc.

It could get complicated if shrapnel can cause damage, but computers should be consistant, so if player1 fires a rocket at a truck, all players on the server should end up with wheels landing in the same spot. However, this does give complications when people dont have the phsyics processing capacity to work the stuff out -> so i suspect that damage from shrapnel wont be around for a while in multiplayer games.

So, imo, I dont see why having phsyics enabled on a system will increase the bandwidth to any extent to cause problems. Obviously objects will carry more information, but again, I dont see why this will increase bandwidth as the data can still all be client side with the exception of actions (players firing, moving etc).

I could be completely wrong, but I think it sounds right