Early Fermi Benchmarks

[YeuEmMaiMai]

problem with physics done on the GPU (any version) is that it will degrade video rendering performance...only way that you could get a boost is to limit the frame rate or dumb down the graphics so that the video card actually has TIME to do the physics work.

 

[GoldenTiger]

problem with physics done on the GPU (any version) is that it will degrade video rendering performance...only way that you could get a boost is to limit the frame rate or dumb down the graphics so that the video card actually has TIME to do the physics work.

Yes, and somehow if it's done on the CPU it magically doesn't cost any performance on that front, you're saying? Amazing, where do you get this performance from thin air from? This magical technology is relevant to my interests!

/sarcasm for those who are sarcasm-impaired

 

[YeuEmMaiMai]

Yes, and somehow if it's done on the CPU it magically doesn't cost any performance on that front, you're saying? Amazing, where do you get this performance from thin air from? This magical technology is relevant to my interests!

/sarcasm for those who are sarcasm-impaired

well on the CPU side it could (well most likely) have an impact (if it is not coded to use multiple cores for example) There are a lot of games that do not use a dual core let a lone a quad core's full potential....

my point was for the people claiming that physics on a GPU was a godsend to mankind and that just isn't the case

 

[PrincessFrosty]

With that said, anybody claiming that Fermi must be MUCH faster than ATI's offering because its 6 months late simply does not understand how those cards are developed. Nvidia has been busy finalizing the product they wanted to release back in November, this delay time does *not* translate to extra perf (except, perhaps, a bit on the driver side).

No one cares how they're developed, it's not important to our buying choices. What matters is that tech gets predictably better over time and we expect to see newer tech that is faster than old tech, this heavily factors into our buying choices.

If for example I knew for a hard fact that Fermi was going to be 7-8 months late and only say 10-15% faster, and I knew this back when the 5870 launched then I'm not going to be waiting 7-8 months for only 10% extra performance when I can be powering my 2560x1600 display at max res for those 7-8 months.

So yes, the huge delay sucks for nvidia, but I (and anybody looking for a GPU upgrade after March 26), would still buy it if its a better offering all things considered. It doesn't need to be "50% better" or "10% better per month delayed". That is nonsense FUD.

Sure if Fermi releases shortly and its approximately the same price but faster, anyone simply wanting a new video card upgrade right now would sensibly pick the better card.

How you judge how good a card is, is really function of what time it was released in, there is an expected technology growth and some hardware sits above that and some sits below it...many of us aren't forced into buying upgrades, we often wait for cards that are especially good for their time and invest in that, sometimes its worth the wait.

Fermi for it's time doesn't look like it's going to be very good, mostly in part due to delays sure, but I dont CARE why, only that it is. Lets call it a good 6-8 months late, with tech reliably doubling in power every 18-24 months I would say it's 1/3 of a generation behind so I would expect 1/3rd of a generation speed improvement for waiting that time otherwise as a buying choice it's not as good value for money, because buying choice heavily factors in the time you buy it.

If Nvidia want me to wait 6-8 months for me to buy their card instead of an ATI one then they better damn well provide 6-8 months worth of extra performance or no sale.

 

[Raap]

Yes, and somehow if it's done on the CPU it magically doesn't cost any performance on that front, you're saying? Amazing, where do you get this performance from thin air from? This magical technology is relevant to my interests!

The point would likely be to actually utilize the CPU; the development of multi-core CPUs is far ahead of what games can take advantage of, so you might as well use the processing power for something. Meanwhile, our video cards are still struggling with the best-looking games at high resolutions, so lowering polygons/shaders/texture quality( to keep the performance acceptable ) just to tack on some fancy PhysX effects...Well, you get the idea.

 

[Mr. K6]

I answered, the fact it doesn't fit your realm of unrealism isn't my problem. The fact remains, it was slower than 8800GTS's and was a huge fricken flop. Hell, the 2900XT review by Kyle started a thread at B3D over making graphs from fraps while actually playing the games or using canned benches like everyone else, including the site you linked to. And Kyle proved canned benches DO NOT represent actual game play frame rates.

So by all means, stick to your canned benches if it makes you feel better about a shitty product, the rest of use will continue to use [H] benchmarking style as the baseline for which all revies are judged.

And once again more crying that doesn't negate a single thing I said. They tested five games on one system, that's really definitive. Of course, when you're a butthurt fanboy, logic doesn't have any place in your arguments. So run away with your tail between your legs and stop wasting space in my thread with your stupidity.

He's talking physics. You're talking pre-scripted. HUGE difference there.

That's the funny thing, there's not (at least in the final product). And it's a point you PhysX fanboys seem to constantly try to gloss over and hide.

 

[Digital Viper-X-]

all physics are pre-scripted, Physics are nothing but calculations, if you shoot the wall the same way every time, you'll get the same result every time.

But there is a big difference between GPU Physx for NV, and physics done say, in BF BC2 and right now it's a clear win for BF BC2. The destructible world in BF BC2, is interactive, as in a collapsing house or building can kill your enemy, can that be done with GPU Physx? No, because right now it's limited to making things pretty.

 

[GoldenTiger]

And once again more crying that doesn't negate a single thing I said. They tested five games on one system, that's really definitive. Of course, when you're a butthurt fanboy, logic doesn't have any place in your arguments. So run away with your tail between your legs and stop wasting space in my thread with your stupidity.
That's the funny thing, there's not (at least in the final product). And it's a point you PhysX fanboys seem to constantly try to gloss over and hide.

Yes, because we disagree with your almighty opinion, we are "butthurt fanboys" who cannot use logic and are Physx trolls "wasting space in my thread with your stupidity".

 

[GoldenTiger]

all physics are pre-scripted, Physics are nothing but calculations, if you shoot the wall the same way every time, you'll get the same result every time..

When I say pre-scripted I mean "if you shoot the wall, it will break the same way every time." When I say GPU PhysX/Havok, I mean "it will break depending on where you shot it and with what force it was attacked, in a dynamic way based on physics formulas." Your argument is akin to arguing that all video is pre-scripted on computers, by implying that having calculations means it's all pre-done. It doesn't.

 

[razor1]

And once again more crying that doesn't negate a single thing I said. They tested five games on one system, that's really definitive. Of course, when you're a butthurt fanboy, logic doesn't have any place in your arguments. So run away with your tail between your legs and stop wasting space in my thread with your stupidity.

Hmm it was a givin that the r600's performance with AA was poor, it could only do 2 samples per clock vs the g80 4 samples per clock and the r600's AA performance was worse then the r580 and thats what caused its performance to be less then or around the 8800 gts.

 

[Digital Viper-X-]

When I say pre-scripted I mean "if you shoot the wall, it will break the same way every time." When I say GPU PhysX/Havok, I mean "it will break depending on where you shot it and with what force it was attacked, in a dynamic way based on physics formulas." Your argument is akin to arguing that all video is pre-scripted on computers, by implying that having calculations means it's all pre-done. It doesn't.

it's pre-determined, regardless of type, physics will always repeat itself of the conditions are the same, my argument wasn't that all the video is pre-rendered either, it's that certain reactions can be estimated with enough accuracy that you can "pre-render" them your argument that physx / havok is dynamic is wrong for 2 reasons,
1) there is no material density involved, no force, no REAL physics being done, the only thing that is being calculated is he location of the shot in that scenario
2) it's visual, anything visual isn't really getting us any further in game play, then DX10 or DX11 did,

EITHER WAY, Physx is not providing an interactive world, this is what it should be working on, just like all of those cool tech demos were showing. None of that has been implemented.

 

[XMAN245]

And once again more crying that doesn't negate a single thing I said. They tested five games on one system, that's really definitive. Of course, when you're a butthurt fanboy, logic doesn't have any place in your arguments. So run away with your tail between your legs and stop wasting space in my thread with your stupidity.
That's the funny thing, there's not (at least in the final product). And it's a point you PhysX fanboys seem to constantly try to gloss over and hide.

Glad I don't live in your alternate universe.

And as to physics in BC2, it is prescripted. No matter what you use to bring a building down, it ALWAYS looks the same. Boring halfassed crap.

it's pre-determined, regardless of type, physics will always repeat itself of the conditions are the same, my argument wasn't that all the video is pre-rendered either, it's that certain reactions can be estimated with enough accuracy that you can "pre-render" them your argument that physx / havok is dynamic is wrong for 2 reasons,
1) there is no material density involved, no force, no REAL physics being done, the only thing that is being calculated is he location of the shot in that scenario
2) it's visual, anything visual isn't really getting us any further in game play, then DX10 or DX11 did,

EITHER WAY, Physx is not providing an interactive world, this is what it should be working on, just like all of those cool tech demos were showing. None of that has been implemented.

I'm guessing you never play GRAW2 or any other game with reactionary physics put to use cause if you had, you'de be able to see that BC2 uses the same thing over and over and over again, while games that use reactionary based physics have multiple different effects based on how and from where you interact with the object. Huge difference.

 

[XMAN245]

Hmm it was a givin that the r600's performance with AA was poor, it could only do 2 samples per clock vs the g80 4 samples per clock and the r600's AA performance was worse then the r580 and thats what caused its performance to be less then or around the 8800 gts.

Can't reason with someone living in an alternate universe.

 

[BravO)))]

Glad I don't live in your alternate universe.

And as to physics in BC2, it is prescripted. No matter what you use to bring a building down, it ALWAYS looks the same. Boring halfassed crap.



I'm guessing you never play GRAW2 or any other game with reactionary physics put to use cause if you had, you'de be able to see that BC2 uses the same thing over and over and over again, while games that use reactionary based physics have multiple different effects based on how and from where you interact with the object. Huge difference.

So how much did physx effect the environmental gameplay in GRAW2?

 

[Digital Viper-X-]

Glad I don't live in your alternate universe.

And as to physics in BC2, it is prescripted. No matter what you use to bring a building down, it ALWAYS looks the same. Boring halfassed crap.



I'm guessing you never play GRAW2 or any other game with reactionary physics put to use cause if you had, you'de be able to see that BC2 uses the same thing over and over and over again, while games that use reactionary based physics have multiple different effects based on how and from where you interact with the object. Huge difference.

Ugh, when did I say no other game does this? I'm saying GPU Physx doesn't do this, it just prettyfies things

 

[XMAN245]

So how much did physx effect the environmental gameplay in GRAW2?

I found it made it more fun as you could shut throw and destroy certain objects in the game where as without the patch, things like fences provided protection until it was knocked down completely.

 

[XMAN245]

Ugh, when did I say no other game does this? I'm saying GPU Physx doesn't do this, it just prettyfies things

Havnig played GRAW2 and DARKVOID, it does alot more than prettyifies the game as it gives it a more realistic feal as opposed to prescritped same crap all the time based physics. In DARKVOID, ricochets and debris from ronds actually cause your character to go ouch as you get pelletd with them.

 

[Digital Viper-X-]

Havnig played GRAW2 and DARKVOID, it does alot more than prettyifies the game as it gives it a more realistic feal as opposed to prescritped same crap all the time based physics. In DARKVOID, ricochets and debris from ronds actually cause your character to go ouch as you get pelletd with them.

yea, you can also ricochet bullets in other games like MW2.. wow, it's not the GPU physx that causes that by the way, it is physx, but done on the CPU, not GPU. I find the "pre-scripted" crap you keep talking about, makes a game much more fun then some fog and paper on the ground

 

[jaredpace]

makes a game much more fun then some fog and paper on the ground

I got really pissed when they took those two effects out of Batman AA when running CPU physx. Nearly every special effect was gpu-only Physx effect in that game, unless you edited the config files. I mean papers & fog physics in particular are run on the cpu in multiple pc titles.

 

[XMAN245]

yea, you can also ricochet bullets in other games like MW2.. wow, it's not the GPU physx that causes that by the way, it is physx, but done on the CPU, not GPU. I find the "pre-scripted" crap you keep talking about, makes a game much more fun then some fog and paper on the ground

DARKVOID is GPU PhysX and the ricochets cause pieces of rock to fly about aswell. Again, the PhysX in DARKVOID is not for the prettier affect in the game.

 

[Digital Viper-X-]

DARKVOID is GPU PhysX and the ricochets cause pieces of rock to fly about aswell. Again, the PhysX in DARKVOID is not for the prettier affect in the game.

Correct me if I'm wrong, but I remember an article stating that currently GPU physx are limited to special effects. because the game cannot send data to the gpu -> back to the cpu -> back to the gpu(Developers are not willing to do this as it will limit game functionality to NV only, and half their sales(xbox / ps3, ATI / NV)

Physx can be done on the CPU / GPU at the same time, and you don't need an NV GPU to run cpu physx.

To further drive the point home for you

There is no doubt about it, PhysX makes a tremendous visual difference in Dark Void. The advanced particle effects enabled by PhysX make land combat an exciting and richly detailed experience. Between the shards of broken enemies, the dust clouds from bullet strikes, and the awesome torrent of particle streams from the Disintegrator cannon, combat in Dark Void is just plain awesome with NVIDIA’s PhysX technology running the show. Even Will’s jetpack is made more awesome with PhysX. That isn’t to say that the jetpack wasn’t cool without it, but with PhysX running on High, the stream of smoke coming out of the little jet engines is thick and dense and cool.

♠end of the day, what we are talking about here is still nothing more than some eye candy. PhysX is not a game-changer for Dark Void. It makes the game prettier, but it doesn’t actually change the way the game is played. That is, of course, logical. If Airtight had made Dark Void’s gameplay mechanics depend on the presence of PhysX acceleration, there would be a few limitations in place. First, they would potentially cut themselves off from a rather large swath of customer with AMD’s ATI Radeon graphics cards. If they didn’t want to do that, they would be forced to lower the complexity of their PhysX-based content so that it could be effectively calculated on the CPU.

http://www.gamephys.com/2010/03/13/...rysis-2-with-awesome-physics-and-destruction/
for fun =p

 

[razor1]

Correct me if I'm wrong, but I remember an article stating that currently GPU physx are limited to special effects. because the game cannot send data to the gpu -> back to the cpu -> back to the gpu(Developers are not willing to do this as it will limit game functionality to NV only, and half their sales(xbox / ps3, ATI / NV)

Physx can be done on the CPU / GPU at the same time, and you don't need an NV GPU to run cpu physx.

Depends on the type of physics, some game changing physics will be limited to only GPU based because of this, it has nothing to do with vendor specific it has to do with limitations of the hardware itself. The data can be sent back and forth but the bandwidth needed will cause problems, but it all depends on what you are going for. Like this, if you want a tesselated object to break down accurately like in real life there is going to be problems with that, mesh density is going to cause a huge bandwidth problem. So things like what Xman just wrote, with using partciles that are generated on the fly is a much better option. The more you want real world feel for physics, game design is still limited to the base hardware we have. Until those limits are removed, real world physics will not be possible. This is why the CPU is going to be a crutch. As it is right now its already becoming a crutch. But again, to do things like this, the game design has to take into concideration of physics right off the bat. So games that are starting off with production goals when physX was released for nV gpus (a year and a half ago or so) are the games we will see physX being used more in game changing rolls, like DarkVoid and Metro. Early games GPU physX was just an add on, they might have been using the physX engine but game design wasn't made with it in mind.

 

[YeuEmMaiMai]

thread title misleading thread title......lol how can there be early benchmarks of an already 6 month late product?

 

[noquarter]

it's pre-determined, regardless of type, physics will always repeat itself of the conditions are the same, my argument wasn't that all the video is pre-rendered either, it's that certain reactions can be estimated with enough accuracy that you can "pre-render" them your argument that physx / havok is dynamic is wrong for 2 reasons,
1) there is no material density involved, no force, no REAL physics being done, the only thing that is being calculated is he location of the shot in that scenario
2) it's visual, anything visual isn't really getting us any further in game play, then DX10 or DX11 did,

EITHER WAY, Physx is not providing an interactive world, this is what it should be working on, just like all of those cool tech demos were showing. None of that has been implemented.

The outcome is predetermined, in a cosmos sense. In programming terms scripted has a very specific meaning - if something is scripted you don't have to solve the equation, the outcome is set at the start. If it's calculated you have to determine the outcome through math, even if the outcome is the same at the end you still have to solve for it, which takes much more work.

It's like when your math teacher told you to 'show your work'. Scripted means you looked up the answer and wrote it down without doing the work.. same outcome, but no work involved.

The difference ends up being a pre-scripted event has very limited interactibility, while solving the equations through a physics API allows for complete interactibility since your actions are the input values.

 

[Digital Viper-X-]

The outcome is predetermined, in a cosmos sense. In programming terms scripted has a very specific meaning - if something is scripted you don't have to solve the equation, the outcome is set at the start. If it's calculated you have to determine the outcome through math, even if the outcome is the same at the end you still have to solve for it, which takes much more work.

It's like when your math teacher told you to 'show your work'. Scripted means you looked up the answer and wrote it down without doing the work.. same outcome, but no work involved.

The difference ends up being a pre-scripted event has very limited interactibility, while solving the equations through a physics API allows for complete interactibility since your actions are the input values.

Sorry, I know the difference =) I was just being a pest, since a lot of fanboys are grasping at the "pre-scripted crap" as opposed to the "visual only" crap physx puts out. I'm splittin hairs!

 

[WabeWalker]

I imagine that by now you guys have seen the Metro 2033 benchmarks - in DX 11, at 1920x1080 the ATI cards appear to be struggling with tessellation. This is especially true when 4xAA is enabled, and when depth-of-field is turned on.

In that one benchmark for the 5850 it registered 11 frames per second.

It'll be interesting to see what happens when the GTX 480 is released - will it tessellate better, and will it actually make a difference in the visuals? Obviously there's no point in having better tessellation results if you can't actually make out the difference while playing the game. The numbers aren't always everything.

I'm on the fence right now - should I get a 5970 if I can find one, or should I buy the single GPU GTX 480? My GTX 275 is performing amazingly well, but I'd like to try to stay ahead of the curve this time out. Tough decision here.

I think it's probably more prudent to wait for some concrete GTX 480 reviews before deciding.

 

[shansoft]

I imagine that by now you guys have seen the Metro 2033 benchmarks - in DX 11, at 1920x1080 the ATI cards appear to be struggling with tessellation. This is especially true when 4xAA is enabled, and when depth-of-field is turned on.

In that one benchmark for the 5850 it registered 11 frames per second.

It'll be interesting to see what happens when the GTX 480 is released - will it tessellate better, and will it actually make a difference in the visuals? Obviously there's no point in having better tessellation results if you can't actually make out the difference while playing the game. The numbers aren't always everything.

I'm on the fence right now - should I get a 5970 if I can find one, or should I buy the single GPU GTX 480? My GTX 275 is performing amazingly well, but I'd like to try to stay ahead of the curve this time out. Tough decision here.

I think it's probably more prudent to wait for some concrete GTX 480 reviews before deciding.

I really dont see any difference in this game with tessellation on

can anyone tell a difference?

 

[deff]

I imagine that by now you guys have seen the Metro 2033 benchmarks - in DX 11, at 1920x1080 the ATI cards appear to be struggling with tessellation. This is especially true when 4xAA is enabled, and when depth-of-field is turned on.

In that one benchmark for the 5850 it registered 11 frames per second.

It'll be interesting to see what happens when the GTX 480 is released - will it tessellate better, and will it actually make a difference in the visuals? Obviously there's no point in having better tessellation results if you can't actually make out the difference while playing the game. The numbers aren't always everything.

I'm on the fence right now - should I get a 5970 if I can find one, or should I buy the single GPU GTX 480? My GTX 275 is performing amazingly well, but I'd like to try to stay ahead of the curve this time out. Tough decision here.

I think it's probably more prudent to wait for some concrete GTX 480 reviews before deciding.

From looking at the "Optimal System Specs" this game looks like its going to be the next Crysis. Thats not a bad thing, but it's going to need a lot of horsepower to run at maxed settings.

* Core i7 CPU
* NVIDIA DirectX 11 compliant graphics card (GeForce GTX 480 and 470)
* As much RAM as possible (8GB+)
* Fast HDD or SSD

Also, do you have a link to these benchmarks?

 

[Mr. K6]

Yes, because we disagree with your almighty opinion, we are "butthurt fanboys" who cannot use logic and are Physx trolls "wasting space in my thread with your stupidity".

Stop making inane arguments and I'll stop calling you out on them. It's that simple.

Hmm it was a givin that the r600's performance with AA was poor, it could only do 2 samples per clock vs the g80 4 samples per clock and the r600's AA performance was worse then the r580 and thats what caused its performance to be less then or around the 8800 gts.

The card was unbalanced, no doubt (512-bit ring bus?). The shoddy drivers didn't help it either. But people going on and on about it (while rewriting history), especially in a thread about Fermi, is ridiculous.

Glad I don't live in your alternate universe.

Yes, please run from the discussion before you're embarrassed further. Thank you.

And as to physics in BC2, it is prescripted. No matter what you use to bring a building down, it ALWAYS looks the same. Boring halfassed crap.

And if BC2 did the exact same thing with PhysX you'd be wetting yourself and liken it to the second coming of the video gaming Christ. But somehow Havok, using only one game, completely showed up anything PhysX has done in the last three years. So now you're going to have a temper tantrum and cry and try to shoot it down in any thread it's brought up in because it completely wrecked your beloved NVIDIA/PhysX/whatever you worship. The truth of the matter is, no one cares about scripting because scripting mimics real world physics 99% of the time. Dynamic means shit, especially in a multiplayer game. No one is standing around to see if a piece of wood falls differently each time the house collapses.

I really dont see any difference in this game with tessellation on

can anyone tell a difference?

Considering it's pretty much the same engine as XRay, yeah, that's about all you get out of it. Pouches on clothing look more puffy and helmets are slightly smoother. It's just like that in STALKER: Call of Pripyat, and very difficult to pick out in gameplay. What I would like to see is those benchmarks, especially because my 5870 has zero issues running DX11 and Tessellation in STALKER: CoP.

 

[LeninGHOLA]

I imagine that by now you guys have seen the Metro 2033 benchmarks - in DX 11, at 1920x1080 the ATI cards appear to be struggling with tessellation. This is especially true when 4xAA is enabled, and when depth-of-field is turned on.

In that one benchmark for the 5850 it registered 11 frames per second.

It'll be interesting to see what happens when the GTX 480 is released - will it tessellate better, and will it actually make a difference in the visuals? Obviously there's no point in having better tessellation results if you can't actually make out the difference while playing the game. The numbers aren't always everything.

I'm on the fence right now - should I get a 5970 if I can find one, or should I buy the single GPU GTX 480? My GTX 275 is performing amazingly well, but I'd like to try to stay ahead of the curve this time out. Tough decision here.

I think it's probably more prudent to wait for some concrete GTX 480 reviews before deciding.

I'd be more worried about how well optimized the game is rather than how well video cards can play it right now.

 

[WabeWalker]

From looking at the "Optimal System Specs" this game looks like its going to be the next Crysis. Thats not a bad thing, but it's going to need a lot of horsepower to run at maxed settings.



Also, do you have a link to these benchmarks?

Yes, right here. This was posted by another forum member in the PC Gaming section of this forum. You guys should visit that section. It gets pretty crazy over there. I'd say that the two 'hot spots' on the forum are here, in the GPU section, and over there in the gaming section.

I find that people tend to take their games and their GPUs damned seriously. Damned seriously. Although I also have a working theory that the people are mostly guys. I don't believe that there are too many females posting here.

http://www.pcgameshardware.de/aid,7...tX-11-und-GPU-PhysX/Action-Spiel/Test/?page=2

 

[WabeWalker]

Shansoft, I can't see any difference at all. Which one is tessellated? The top one looks slightly darker, but that's about it. They look like the exact same image to me?

 

[shansoft]

Shansoft, I can't see any difference at all. Which one is tessellated? The top one looks slightly darker, but that's about it. They look like the exact same image to me?

second one is with tessellation....

 

[FranklyShankly]

Shansoft, I can't see any difference at all. Which one is tessellated? The top one looks slightly darker, but that's about it. They look like the exact same image to me?

It is different. if you look closely you can see in the tessellated image the pocket on the sleeve is slightly larger, the wrinkles in the cloth a little deeper, the bars on the window a little thicker. it's slightly different, but does it look any better? Not really imo.

 

[Infaustium]

It is different. if you look closely you can see in the tessellated image the pocket on the sleeve is slightly larger, the wrinkles in the cloth a little deeper, the bars on the window a little thicker. it's slightly different, but does it look any better? Not really imo.

Have to agree, not really any significant difference... and that's coming from someone who is definitely big on visuals.

 

[MrLonghair]

I can tell a difference between the shots. The tessellation looks like shit put on everything. As if someone ordered it set to maximum. Easy to see we got the bullshot treatment of the game's hype-period.

Metro2033 surprised me regarding its PhysX implementation. Either it's got nearly nothing that uses it or this is the only developer to do more than simply jam PhysX code into their game - about as much performance loss when running on the CPU as with any other physics system in a game - barely anything! If only that could apply to the other PhysX games, or you know, the PhysX GPU code? Maybe it's a sign that we don't need extra hardware, we only need fine people on the developing team?

(PhysX is also used for effects, particles etc, in case someone didn't know. The fact that it runs so well on CPU in Metro2033 has my thumbs up, although there are wide-screen issues on many systems, and forget about eyefinity (I'm told.))

i c wat u did thar.

<3

 

[FranklyShankly]

I can tell a difference between the shots. The tessellation looks like shit put on everything. As if someone ordered it set to maximum. Easy to see we got the bullshot treatment of the game's hype-period.

Metro2033 surprised me regarding its PhysX implementation. Either it's got nearly nothing that uses it or this is the only developer to do more than simply jam PhysX code into their game - about as much performance loss when running on the CPU as with any other physics system in a game - barely anything! If only that could apply to the other PhysX games, or you know, the PhysX GPU code? Maybe it's a sign that we don't need extra hardware, we only need fine people on the developing team?

(PhysX is also used for effects, particles etc, in case someone didn't know. The fact that it runs so well on CPU in Metro2033 has my thumbs up, although there are wide-screen issues on many systems, and forget about eyefinity (I'm told.))

i c wat u did thar.

 

[XMAN245]

I didn't say it was a good card, I said it was fine. It didn't suck, it excelled in some areas, and tanked in others. Through all of your hissy fits you haven't shown otherwise anything besides you get your panties in a bunch over video cards. Take the nerd rage down a notch.
Let me guess, you never owned, worked, or even saw one, but somehow can say this definitively from reading a review on [H] that tested it in only five games. I don't know what's worse, that you're that stupid or that arrogant, either way, you lose.

The word you were looking to use is "deluded;" don't use big words if you can't even spell them . Like I said, never used it, read a review, thinks he knows something - hilarious.

So you can't even defend PhysX anymore so you'll try to bring down Havok to its level. Interesting tactic, is this the "at least it sucks as much as PhysX argument?" Kind of along the lines of "at least it sucks as much as the HD2900XT?"

Why does there need to be one? Is this the "at least we have one argument?" How many ways are you going to try to deflect PhysX's failures before you just suck it up and go cry in the corner?


*deafening . Again, don't use big words if you can't even spell them (which is ironic, because that isn't even a big word, yet you still can't spell, quite sad really).

Actually I did own one for about 3 months, 9 months after launch. In every game I played, IT SUCKED compared to the 8800GTS I also owned. that covers about 25+ games, each and every single one of them performaned BETTER on the 8800GTS than on the 2900XT. I'd call tha,t as would anyone else with atleast 1 functional brain cell let alone a few billion, a card that sucked, clearly you dont fit the bill.

In the realm of physics APIs, those that matter anyways, CryTek leads the way, next is PhysX and last and pulling up the rear buy a far margin is Haovk. It is not deflecting anything, it is being able to admit there is something out there better than PhysX and the CryTek engine is it. Havok is in a very distant third when the three are compared. And no, I meant Deluted, as you are surely dellusional to think the 2900XT wasn't a flop. It was the stinkiest fish in the fish market after have been left out in the sun for 4 days surrounded by fresh catches

As far as who is worse, it has to be you. It takes a very sad person to call Kyle and his reviewing style poor and not very good. How you can stay here is beyond me.

And as to my spelling, I never claimed to be a spelling be champion, let alone spelling bee anything.

 

[CommanderFrank]

Get back on topic. Next one who slings insults will ghet a nice long vacation to think it over.

 

[Trimlock]

world record, 6 deleted posts in a row about freakin PHYSX?!?! no one in their right mind cares about physx right now.

why don't they care about physx? because Nvidia locked it down with their proprietary bull shit, look to see how it pissed off kyle with one of his recent articles about freaking out on them if they decide to lock one of their TWIMBP games out of eyefinity.