Not much more I can say on this - check our Rys' overview http://www.hexus.net/content/item.php?item=5492
Not much more I can say on this - check our Rys' overview http://www.hexus.net/content/item.php?item=5492
Meh
Interesting; basically, at the moment, then, you're paying £200 for a card that actually causes more work for one of the more expensive bits of your system, i.e. your graphics card, with a performance penalty, in order to get some more blobby bits that vanish within a few seconds anyway. I'm not seeing an upside here...
I note that Havok have been having a go; http://www.firingsquad.com/news/news...searchid=10096 . OK, they're in competition, effectively, but from what you describe, they have a point...
Some interesting posts on this forum about the topic.
http://forums.overclockers.co.uk/sho...php?t=17568825
There are some videos in there aswell which show the difference between normal and physX enabled gameplay.
Personally I dont think there is a great deal of difference looking at the videos and I never would have expected frame rates to drop so much.
U'd probably be better off spending £200 on a 2nd card for an SLi system than this.....
Can't help but agree; more blobs or higher res with better frame rates...not exactly a tough call .
Looks like a really stupid waste of money right now to me.. I think the ideas got lots of mileage but i'd be suprised if Ageia stay at the forefront for it. ATI and nVidia are making physics-related noises now and microsoft are talking about a directx-type API for it (which will no doubt become the defacto standard very quickly).
Perhaps the sick truth for now is that you'd be better off with dual core - it's more common so software developers would be more likely to offload physics processing to a second core versus a specialist 200quid card with low market penetration.
you mention
Are there any modelling programs that support it yet? What sort of performance increase would be seen and in what areas of the modelling?There's support in the API for limited fluid dynamics sims, vehicles (wheel, torque and tire simulation), object raycasting and more, which the PPU can fully or partly accelerate (with Ageia moving more onto the hardware as time goes by).
your final point about fragmented APIs gave me a huge deja vu attack.
3dfxs Glide vs MS' Direct3D?
People said virtually the same for Voodoo. (Get a faster CPU!, wait for the next Intel!) Blah Blah. Waste of time! No games! *GLQuake* Ooooh cute. I wants one! *robs a granny*Originally Posted by dangel
hehe.
Deja Vu
DeathByDuke is probably right.
If PhysX can get enough developers to support their PPU (exclusively would be the clincher...) they'd be on a winner.
That requires a nice API and by the sounds of it, theirs needs some polishing.
I'm going to wait and see what Unreal and other engines produce. I think Ageia should wait for the big launch push until they have some impressive titles
So you'd be better off with an older/slower cpu and a ageia card for the next six months? Can't quite see it your way GPU and CPU will be all important for at _least_ that period.Originally Posted by DeathByDuke
And therein lies the problem - if someone like Microsoft/ATI/nVidia/Havok do something then it's going to make life very difficult for Ageia. I think as an end-user i'd rather see MS make a directX physics API so that we can choose whatever hardware methodology we like to render it (be it cpu, gpu or ppu) and as a developer i'd be much more comfortable with that in terms of meeting the market.Originally Posted by Steve
I have to agree with dangel that we need a common API, like OpenPL or DirectPhysics. It's kind of like blue-ray and hd-dvd in that we need a standard, I mean do you want to end up with two PPUs for different games, seems like a waste of money to me.
As for a slight preformance hit, its worth it if it can do a much better job renering, like it's worth the hit for AA, but until game developers make games that really take advantage of it that won't happen and we are likely to have a chicken and the egg problem with develpers not waanting to write 2 versions or rely on software PPU for extened fancy stuff.
The heat from device makes me wonder how long until people start putting aftermarket cooling on these things, like a Zalman device or a liquid cooling sytem. That in turn makes me wonder how long until people figure out how to overclock these things just to say they did.
Just a thought, but I think they should have made the card PCI Express instead of PCI.
After all, it is not going to go into any legacy systems, and all new systems have spare PCI-Express slots that are just sitting there doing nothing.
There are currently 1 users browsing this thread. (0 members and 1 guests)