Cellfactor and Physx

[Giftmacher]

I didn’t say Tomshardware is rubbish I said its gone downhill.

Oh I see, it is a good site Let's just say that isn't what comes across in the OP, especially when you accuse them of making amateurish mistakes.

Anyway I read the review ages ago when it came out I think, the GRAW one? I don't really see the bearing on the updated review, which did point out valid problems at the time. Things have moved on, that doesn't devalue the work done then it just means it no longer applies. As I've said already Physx needs a decent title that really shows it's capabilities off before anyone judges it, Cellfactor gives a nice flavour but this hardware needs more than a tech demo to get market penetration.

Gift.

 

[james.miller]

tomshardware is a terrible site, it always has been. they are haevily biased towards whoever happened to pay them more money. their review are often wrong and they make no attempt to correct mistakes. Im not even sure how anybody can argue otherwise lol

 

[Pottsey]

“Oh I see, it is a good site Let's just say that isn't what comes across in the OP, especially when you accuse them of making amateurish mistakes..”
You say accuse like they didn’t do anything wrong! My words where they made mistake that even armatures wouldn’t make which is true and I stand by what I said. Just go read what they wrote.

But sorry if it came across as the whole site was rubbish. I didn’t mean that. Some parts of it are good. Only the PPU reviews was worse then amateurish and the odd 3dcard review which are poor but better then amaterus (not all 3dcards reviews are poor, most are ok)





“which did point out valid problems at the time. Things have moved on, that doesn't devalue the work done then it just means it no longer applies.”
Err the valid problems at the time where not valid that’s my point. There are problems with there old and updated review. Both are done by amateurs. Just look how the benchmarks where done. When benchmarking you don’t run the two things your trying to compare against with vastly different settings.

You don’t run object A on high settings then object B on low settings then go object B is faster don’t buy object A as its slow. That’s just what tomshardware did.

 

[Giftmacher]

But sorry if it came across as the whole site was rubbish. I didn’t mean that. Some parts of it are good.

Ok let's not go on about it then I don't want to keep moaning at you over a misunderstanding. You have my apologies too.

When benchmarking you don’t run the two things your trying to compare against with vastly different settings.

You don’t run object A on high settings then object B on low settings then go object B is faster don’t buy object A as its slow. That’s just what tomshardware did.

Actually I was thinking about the graphical glitches etc. They needed to be pointed out at the time. As for the rest I'd agree but I think the comparison is even more fundamentally flawed than you say. IMO it simply isn't a valid comparison to take code optimised for hardware with a PPU and compare between rigs. That said I've seen loads of people try it and consequently come to all sorts of contrary conclusions. Perhaps when Havok FX puts in a showing we can at least make a stab at comparing competing tech in a meaningful way. However, to be truly comparative that'll still take a common application able to use either method of physics acceleration and I don't see that coming soon (if at all).

Gift.

 

[Pottsey]

To thorw more fuel in the fire Ageia just said this. From another fourm no source link so it might not be true.

“Yes, unfortunately the reason that the physics are not smooth and fluid like in CF: CT is because there was a bug introduced into the PhysX sdk, which then caused the need to reduce the physics accuracy in CF: R to 50%. There was not time to fix this problem before the release date. It may however be fixed in a patch.”

If that’s true can we say incompetence!

 

[V1N]

Just posted my latest review of the PhysX cards. Might show what they can do in a few more games :

Ageia PhysX Review

Ah, I just read, and was about to post that very link.
Came as a link in a gamespy email!

I'm not sure why there's so much negativity towards the PPU.
I'd love to see such a thing take off. Just needs better games support.
I'll be interested to see what difference it makes to Unreal Tournament 3. If they can do something special with it, then it should encourage more people to buy one, and therefore have more games support it in the future.

The whole argument reminds me a little of around 1996/97 when 3dfx released their 3D card. They made things look better, but you didn't need one to play the games, and I remember people telling me what a waste of money it was.

Of course, games improved, and then required a 3D card to be played, and they got better and better. The problem with something like Physx, is that they're restricted to what they can do. At the moment, bolting on a few extra effects for users of the card isn't particularly exciting, and doesn't add anything really new to a game. Imagine if games were still written using software 3d, and you got some extra effects if you had a 3D card, they'd not be able to add an awful lot to it.

If there was a game that required a Physx card, and was written under the assumption that every gaming rig had one, I bet the game would have amazing physx. I remember the first time I loaded up Half Life 2, and spent a while playing in that kids park near the beginning, with the swings and roundabout etc. If a game came out requiring Physx hardware, imagine what it would be like. Racing games would be transformed... I imagine a bit like Flatout 2 or Motorstorm, but with "even more" Or shooters where everything really did interact with everything else in a realistic manner... where everything was destructable, and collapsed realistically.

So I personally do hope it finally takes off.
That's not to say I'm ruling out other solutions either.
If they can do the same job using graphics cards, then great.
I'd love to see the "next-gen" (don't you hate that phrase?) of games coming out, all of which have excellent physics, making the game that much more immersive, make it feel more real, and no more running into a small object, that has the same effect as running into a brick wall.

Two things would have to be changed for me to buy one though.
(1) that review said it has a noisy fan... that's a nono
(2) I have no PCI slots free, but plenty of PCI-E.

Sorry for boring you to death.
Just rambling on during my first coffee of the day... which has now gone cold!

So, hoping we get more physics in the future, and we can enjoy even more action in our frames, whilst maintaining a decent framerate

Just thought I'd say it.

V1N.

 

[Shiz]

Hi V1n, thanks for the kind words. Can you forward me the email? I don't remember sending a link to gamespy so would love to see how they found it and may even ask what made them include it .

Edit: Email in trust.

 

[Chronictank]

No game out currently makes use of proper fluids and cloth simulation. Fluid dynamics and cloth solvers require a lot of resource and theres no game which looks like it is properly making use of how cloth or fluids would be simulated in an industry standard 3d content creation package. Take the application Real Flow for example, http://www.nextlimit.com/realflow/ (this may interest you Pottsey),

If they could incorporate that sort of thigns into games that would definitely make it worth buying a PPU card, as it happens at the moment nothing like that exists using the PPU so i cant see the need to buy one