TheGuruStudI want two gtx+. Lets see you guys top that in performance for $380 (shipping included). That's right, it CAN'T be done. So stfu. GTX+ is an excellent choice if you can't put out for the 260 or 4870.

And framerate drop is negligible from physx accel if you're already getting decent fps. If you're getting such low fps that you need those extra 5 back from the physx, then you need more power anyway.

edit (for guy below :)): I want two of these, specifically, if they don't charge an arm and a leg for the cooler.

p_o_s_pcwow this thread has gotten off topic is becoming a useless argument.:ohwell: but i agree STFU :nutkick:

The 9800GTX+ wouldn't have problems running physX i can run it on my 8800GT without having noticeable frame rate drops

Yeah, well I'm running at 1920x1200. An 8800GT struggles a bit at that res. The difference is noticeable when the card is already being pushed to it's limits. Not in terms of average framerates, but in terms of minimum framerates. When there's nothing for the Physx processor to do, the framerates are identical between both my setups. it's mostly during explosions that the single 8800 falters. Whereas with both cards, there is much less of a frame drop in those situations.

Besides, I never said the GTX+ was a bad card. I was just saying that comparing a 4850 and 8400 combo to a single GTX+ isn't an apple to apples comparison. The 9800GTX+ is automatically at a disadvantage in that situation, as it has to process both, whereas the 4850 doesn't. Never said anything about either being the better buy.

Why are PhysX and CUDA good reasons to choose a NVIDIA card over an ATI card? CUDA isn't even about games, and it's not guaranteed to be useful for much of anything aside from Folding, some future video codecs, or other scientific workstation-stuff. The value there is highly subjective. And, PhysX is a gimmick at this point to say the least. We need a standardized "physics" API instead of this proprietary BS, so everyone will support it.

If PhysX and CUDA actually go somewhere cool, time will have gone by and there will be vastly superior cards to run the apps on. So right now I definitely don't think either tech damages the value of a 48x0 card. Those Radeons have awesome shader power (in excess of even the GTX 2xx cards) that might make them better in the long run, too.

4850 and 9800GTX+ are both nice cards, but I definitely think the 48x0 series is a major win for ATI.

(Oh, and I don't have a Radeon 4xxx card, but a 8800GTX, so no fanboy justification energies here)

swaaye4850 and 9800GTX+ are both nice cards, but I definitely think the 48x0 series is a major win for ATI.

Yeah, it's good for ATI and good for me. Without HD4850 I would have never gotten my 9800GTX+ as it would have been in a completely different price. Competition is great for consumers, as long as the price drops before the purchase :)

swaayeWhy are PhysX and CUDA good reasons to choose a NVIDIA card over an ATI card? CUDA isn't even about games, and it's not guaranteed to be useful for much of anything aside from Folding, some future video codecs, or other scientific workstation-stuff. The value there is highly subjective. And, PhysX is a gimmick at this point to say the least. We need a standardized "physics" API instead of this proprietary BS, so everyone will support it.

If PhysX and CUDA actually go somewhere cool, time will have gone by and there will be vastly superior cards to run the apps on. So right now I definitely don't think either tech damages the value of a 48x0 card. Those Radeons have awesome shader power (in excess of even the GTX 2xx cards) that might make them better in the long run, too.

4850 and 9800GTX+ are both nice cards, but I definitely think the 48x0 series is a major win for ATI.

(Oh, and I don't have a Radeon 4xxx card, but a 8800GTX, so no fanboy justification energies here)

Because:

a) Graphics cards are not only for games. It has been demostrated CUDA can be used to accelerate video transcoding by a factor of 5x-10x and will be used for amazing acceleration on many features in the next Abobe CS4 suit. That alone is a strong selling point for millions of people that don't care a shit about games. Just because there's no apps right now doesn't mean there couldn't be in the future. And there's going to be, many are in development. For someone like me that works with Photoshop and Premiere everyday (I'm waiting for Mental Ray too :respect:), and that loves to transcode many many videos to fit them on my cellphone, CUDA accelerated apps are the Holly Grail. Because gaming performance and price is similar, I see no reason to not prefer and recommend this card over the HD4850.

b) PhysX is far from being a gimmick. If you have tried the demos, Warmonger, UT3, etc. and you still think it's a gimmick, either you don't care about physics or you are not able to understand what's happening in front of your eyes. Again there's no games utilising it NOW, there are going to be around 20 this Christmas and many more in the future. I am the only one that buys the cards with the intention of using them for more than a few months?? Of course not, most people keep them more than 2 years.

c) Off course an open standard physics API would be awesome. Thing is no one will make it. The most "open" one you will see is inside DX11 and that will come in 2010. With the usual adoption rate of new DX versions, we are talking about no HW physics until 2011. Furthermore AMD will not push any hardware accelerated physics until then. If you want hardware accelerated physics before 2010, Nvidia is the only one pushing for it. It's not a matter of Ati can do it or not, because they simply won't do it. As simple as that.

d) Don't be fooled by the peak GFlops numbers Ati advertises. They not only are not useful in games most of the times, but they are far less useful for GPGPU. AFAIK Nvidia cards are much much faster than Ati ones at F@H despite the long time headstart that Ati had to optimize. There's an easy explanation for that, any card by Nvidia since G80 was designed with CUDA in mind, specially the GT200, and that meant a high compromise in performance/price/watt in games. As I said the future of graphics cards is not only gaming. If you want to look only from that point of view, good for you, but NEVER say (or support people that do) one card is not worth it based on your skewed personal opinion. As of me I never said 9800GTX+ is a better deal either, I said that FOR ME, considering similar gaming performance and price, the 9800GTX+ has much more value due to CUDA and PhysX. Again FOR ME.

THE BEST ARTICLE about CUDA you will ever read:

www.realworldtech.com/page.cfm?ArticleID=RWT090808195242&p=1

There you can uderstand why Nvidia designed GT200 for CUDA, with some compromises in detriment of gaming that make it far superior for GPGPU computing.

PhysX is just a novelty like Creative EAX 5. They make the game look/sound better but are optional. As for games that explicitly require PhysX, that's a tough ask in a PC Gaming industry which is already shrinking to console. If a game explicitly asks for NVIDIA hardware in the MSR, that would restrict users of ATI hardware...in essence, lose customers. So I don't see that happening either.

btarunrPhysX is just a novelty like Creative EAX 5. They make the game look/sound better but are optional. As for games that explicitly require PhysX, that's a tough ask in a PC Gaming industry which is already shrinking to console. If a game explicitly asks for NVIDIA hardware in the MSR, that would restrict users of ATI hardware...in essence, lose customers. So I don't see that happening either.

IMHO opinion you can't compare features that only slightly increse quality of one aspect of the game, with something that can change and revolutionate the way we play games. You can't neither compare the number of PhysX enabled hardware base to EAX5 enabled base. Nor you can compare the relationship with developers that Nvidia has with that of Creative.

I know it's hard to happen, but a developer could very well decide to make a game that requires PhysX. By enabling PhysX they can create amazing gameplay features you would not be able to create without. That can be a tremendous advantage in game/developer versus game/deveoper value. It can make a game really stand out of any other game. In the current market making a successfull game has more sense than creating a game that can run on more hardware.
The installed base of next gen consoles is far smaller than PC and overall (all games, only next gen) sales are smaller, but console games like Halo, MGS, GTA are always on top. Why? Because selling to a greater proportion of the installed base has better results than selling to a greater installed base.

Considering the number of PhysX enabled cards are well above 100 million right now and will be doubled soon (most people still have DX9 hardware, once they upgrade it will be natural that 50% of them buy Nvidia=PhysX) and that a PC is really successfull if it sells more than 2 million copies, it could make sense for some developers to concentrate on Nvidia "platform", just like for many developers it makes sense to concentrate on a single console. Usually there's 2-3 games that get the most of the pie, if by using PhysX they think they can be one of those (and in a near future when a 9600GSO like performance is low-end), don't doubt for a moment one or two developers could take that train. And once one of them is successful many will follow. NEVER forget how the GPU entered the scene, for more than a year ONLY Quake could benefit from hardware graphics, and the benefits where smaller IMHO, look where we are today.

What revolutionise? The way you pick up and throw stuff in Half Life 2 / Portal? That's already accomplished by Havoc. My point is, if it's a "NVIDIA only" thing, developers would be foolish to make PhysX a minimum system requirement for a game, at least in a 2008 setting, given that users of ATI hardware (and no additional NVIDIA hardware) wouldn't even buy the game (loss for the developer).

well one thing for sure, Project Torque- altho a free game has physics in it- doesnt require Nvidia (Ageia) Physx, i just believe honestly we need a Universal API and no proprietary BS, aka DirectX/OpenGL/OpenAL/OpenCL are all APIs.

btarunrWhat revolutionise? The way you pick up and throw stuff in Half Life 2 / Portal? That's already accomplished by Havoc. My point is, if it's a "NVIDIA only" thing, developers would be foolish to make PhysX a minimum system requirement for a game, at least in a 2008 setting, given that users of ATI hardware (and no additional NVIDIA hardware) wouldn't even buy the game (loss for the developer).

See? NO! You can do much much more than that with hardware PhysX! It's just waiting there to be implemented. Of course when I talk about the potential of physics I'm not talking about the kind of crappy physics we have today: HL2.

I'm talking about fully destructible environments, much much more than what we can see in Crysis or Warmonger. I'm talking about being able to destroy anything, for example the ceiling above enemies to kill them. ANY ceiling or wall, and them being made of actual bricks. Columns and beams would actually sustain buildings, so if you broke them the building would fall, but only the part of it that should do. And if you shot a rocket to a wall it would make a realistic hole by displacing the bricks that would be displaced in real life, etc. Imagine the fights of COD inside the cities, but the buildings instead of being triggered, everything could be destryed and you would never know if you are actually behind cover, unlike in actual games where you are always safe behind some walls.

I'm talking about warfare in asphyxiating corridors full of fog or smoke where the only way of finding the enemy is because of the displaced fog, while the blow of the guns displace the fog too, opening a door and letting wind enter would disipate it, etc. Much like in FEAR, but without having to wait when there's a lot of smoke, because it's impossible to see anything, even when they are one meter away from you. In real life you would notice them because they displace the smoke.
Or a similar deathmatch, fighting in the middle of sheets or blankets hung on stalls, knowing the path that your prey took, because the sheets are still in movement...

I'm talking about puzzle games where you have to make balls or water (specially I would like one with water) follow a determinate path by using many different pieces, like in those crazy japanese dominoes, etc.

I have many many other ideas, but I would be here forever. I think you can get the idea. I'm definately not talking about throwing a barrel at enemies.

EDIT: As I said if a game really uses physics to it's full potential it could sell more than if they do not implement it EVEN if they only sell to Nvidia owners. 10% of Nvidia owners is much much more than the 2% of both brand owners. Gears of War released only on the XB360, yet it sold more than any other game that year.