Almost lost all faith because of these benchmarks!

[LiquidX]

I have to say the FX from the benchmarks appears to be a card which will only show its full potential if a game is programed directly for it and that will never happen because a no company can risk that with a $400 card.

I almost lost faith however in Nvida because of those benchmarks. I dont mean to go over on the other side but for those who have seen a game called Soul Calibur on the Dreamcast, it was optimzed for the Dreamcast and until this day nothing even on computer beats those grfx. But the the Dreamcast by all accounts died because of the word of mouth not because of its architecture. This same thing is happeing with the FX. We are killing it before we can fully find out what this "CG" thingy is. I have been useing the big "N" since G2 and I will continue this tradition with the FX (maybe) even if it blows up my pc's.

**Sorry for my post always being so long, I guess this explains my no responses:(

[ragejg]

u gotta point tho...

gotta wait a little bit...

drivers have just been too dang good lately... we're all stupified by the recent exposed inefficiencies...

new drivers in the coming weeks, months, whatever... will prolly fix the aa bug and bolster the DX8 code, which of course nvidia couldn't pay much attn to working to *attempt* to perfect the unseen-as-of-yet (sorta) DX9 code...

I think GFFX will mature properly and quickly... no fan fallout needed, just patience *oh yeah i forgot every nv fan has none left!! dang long waiting period!:D*

[LiquidX]

I also must add their is something else to this......there has to be. Nvida has something cooking and it aint just stuffing without no turkey. I predict the big "N" will send out new drivers within a week. To me this was all to cause a commotion. No one is going to tell me that little fairy was for nothing.:nono:



:cry:

[sebazve]

Originally posted by LiquidX
I also must add their is something else to this......there has to be. Nvida has something cooking and it aint just stuffing without no turkey. I predict the big "N" will send out new drivers within a week. To me this was all to cause a commotion. No one is going to tell me that little fairy was for nothing.:nono:



:cry:

hehe wake up no drivers can give 50% speed up unless they were really ****ed up!.:angel:

[LiquidX]

No one is going to tell me that little fairy was for nothing.

[Chalnoth]

Originally posted by sebazve
i hope no one use Cg, just stick to the standard use DX9 and OPENGL HLSL...Cg is bad for us...we dont need propetary crap from any company.
Cg can compile for any (programmable) graphics card, and ATI can write their own back-end to optimize for their hardware. I see no problem with it. In fact, if it is widely-supported, the performance would likely be higher (on any graphics card) than MS's own HLSL. As for the OpenGL HLSL, well, that's not available yet, so it's a mute point.

[sebazve]

Originally posted by Chalnoth
Cg can compile for any (programmable) graphics card, and ATI can write their own back-end to optimize for their hardware. I see no problem with it. In fact, if it is widely-supported, the performance would likely be higher (on any graphics card) than MS's own HLSL. As for the OpenGL HLSL, well, that's not available yet, so it's a mute point.

as far as i know right now Cg only output Nvidia propetary extensions...besides this kind of thing should be in control/involve by various companies like in OpenGL. Its has nothing to do about peformance. Who knows what nvidia might do is this become a standard?

DX is another thing...

[Nv40]

well by looking at 3dmark results.. i think the geforceFx ultra
and a couple of other sintetic benchmarks i think
the Geforcefx has a great potential...

it was more or less than 2x times faster!!! than the radeon9700pro in high polygons counts
and when there where many source lights at same time too...

i mean doesnt Doom3 use heavy use of perpixel lighning
and tons of polys in their levels.. i read somewhere
that every level use from 200k- 350k poligons!!!
thats is more or less 3x times.. the polys of
unreal tournament 2003.. ..

sure .. 3dmark and sintetic benchmarks shows no real
game experience.. but i think they are good for reference..
to know the stronger and weaker points of each card..

someone told something interesting somewhere,
that when DOOM3 ships... we will start again measuring
performance ... like the old times of quake2 when RAW fps
will be more important to decide the winner rather than
memory Bandwidth.. ... and where most bechmarks will
began at 640x480 resolutions.... lol

i mean... all games tested were what? directx7?

would directx9 games will make look a lot better
the Geforcefx?

mmm.. i think the incoming 3dmark in 2003 will have some
interesting info about all this.. not holding my breath ...
but somethings tells me.. that we have not seen all that
the GeforceFx (a pure directx 9 card) is capable to do.

with "pure" i only means that have been designed
and optimized for DIrectx9 in mind since daythe first day ..
like the days if the first release of the Geforce3... it was
slower than the Geforce2ultra in old games and faster in newer
games ..


time will tell..

[The Baron]

Originally posted by goofer456
The arguments against CG have passen many times on this board but here goes:

1) Nvidia controls the language and decided which functionality is supported and which not. This makes Ati and others VPU manudacturars dependent de Nvidia. They will never agree

Would you rather be dependent on nVidia or MS? 'Nough said.

2) The more developers use CG, the more "The way is meant to be played" logo's will appear. Thsi will result in more games not being tested on non NVidia hardware/

Not if other companies aren't stupid and fail to write Cg compilers if it becomes popular.

3) Why use a propriety language as CG if MS HLSL is available?

As Chal said, it's probably faster.

[Chalnoth]

Originally posted by sebazve
as far as i know right now Cg only output Nvidia propetary extensions...besides this kind of thing should be in control/involve by various companies like in OpenGL. Its has nothing to do about peformance. Who knows what nvidia might do is this become a standard?

DX is another thing...
The current list of supported profiles:
* vs_1_1 for DirectX 8 and DirectX 9
* vs_2_0 and vs_2_x for DirectX 9
* ps_1_1, ps_1_2 and ps_1_3 for DirectX 8 and DirectX 9
* ps_2_0 and ps_2_x for DirectX 9
* arbvp1 [OpenGL ARB_vertex_program]
* arbfp1 [OpenGL ARB_fragment_program]
* vp20, vp30 [NV_Vertex_program 1.0 and NV_Vertex_program 2.0]
* fp30 [NV30 OpenGL fragment programs]
* fp20 [NV_register_combiners and NV_Texture_shader)

Notice the ARB_vertex_program and ARB_fragment_program extensions.

[sebazve]

would directx9 games will make look a lot better

who knows and i dont really care, this card will be slow to play dx9 games with all the eye candy, there will be faster cards by the time dx9 games come out

mmm.. i think the incoming 3dmark in 2003 will have some

who cares about this stupid benchmark, it does not represent real world performance besides companies optimize their drivers like hell to look good on this bench so the whole point of this bench is stupid. Right now with today games the gfx has some really important speed issues and IQ problems and thats all i care. It useless to think about future products,games perfomance when we dont no **** about it....

[Demiex]

--------------------------------------------------
hope no one use Cg, just stick to the standard use DX9 and OPENGL HLSL...Cg is bad for us...we dont need propetary crap from any company.
--------------------------------------------------
<shudders at mention of DX> i've had some experiance in that field of programming <shudders again>. Haven't been the same since. Am thinkin bout learn openGL now and direct x programming now scares me after programming a rpg.

But seriously CG is good. Shader programming in DX 8.x was extremely complex from what i've heard, involved some nasty assembly architecture, so a high-level language like CG does nothing but benifit the gaming community. I know the difference in using C++ syntax and using the inline assembler to do calculations, biiiiiiig difference. Who cares if its proprietary, if it produces amazingly details effects with the simplicity of a few function calls, the we all benifit. Game developers can spend less time programmin complex shaders <which are by far goin to be a necessity in future gaming> and more time on gameplay; resulting in faster production cycles, more profit for them, more great games for us. it would be great, though, if they made the language open- source. That would be bad arse, i would have a field day! muhaha

Demiex

[Demiex]

This is interesting too:
"The GeForce FX boasts high precision "studio quality" 128-bit floating point color in all its end-to-end 128-bit pixel processing pipelines glory. This goes beyond DX9's requirements for 128-bit input and output stages and bests ATI's pixel pipeline, which drops down to 96-bit (24-bit per component) for other than some 128-bit texture sampling stages. Ati......answered the call of Carmack for minimun 16-bit fp color support (24-bits preferred), while NVIDIA was matching the challenge tossed out by the film industy."
Quote from Jan. Issue Of Computer Power Users magizine

Seems biased but in reading the whole 20 pg article on various video cards, provides a wide perspecitive. It actually recommended the 9700 pro over the GeforceFX before all this.

hopes someone can use this a bit more and expand on it :)

Demiex

[Filburt]

The precision nVidia offers on the NV30 is certainly a novelty...however, again, the actual power of the chip will hamper it tremendously if 128bit colour precision were actually attempted throughout rendering scenes in the Doom3 engine.

Now, when thinking about 96bit colour...that means we get 24 bits per component (Red, Green, Blue). I seem to remember not long ago in the industry...24bit colour was regarded as the edge of human perceptions. Now, I personally feel this isn't entirely true...however having 24 bits now for each component...I think we're basically reaching the limit of our senses...conscious and subconscious. Having 128bits (36bits per component I guess?) I feel isn't going to make an actual qualitative difference to anyone unless the person comparing the two somehow botches up the 96bit colour rendering. It will, however, perhaps produce quantitative differences...but graphics cards are really about the qualitative experience...not the quantitative. Quantitative is only there to try and benchmark one product against another...and it seems we are slowly moving towards quantitative benchmarking as GPUs become faster and faster.

One thing that *will* likely be qualitatively noticeable with 128bit colour, however, is that it will take an exponentially higher demand on the hardware...which can translate to perceivable performance hit. I honestly don't feel this is a necessary thing considering we likely won't perceive a difference. It suddenly, then, is reduced to only be useful in so-called "bragging rights". I don't buy my hardware for bragging rights, and I think this is certainly a case where this mode of purchase behooves me.

[Demiex]

Nice, Thanks for the constructive post :)

Nite time for me

Demiex

[AGP64]

Originally posted by The Baron
Would you rather be dependent on nVidia or MS? 'Nough said.

If I am a competing VPU manufacturer I am rather dependant on MS than NVidia. They are anyway if you see that DX requirements are determining the features of the VPU's anyway.



Not if other companies aren't stupid and fail to write Cg compilers if it becomes popular..
You're just blind to the reasons other companies are not very willing to support CG. It becomes a different story of MS enters the hardware VPU market.


As Chal said, it's probably faster.

Proof please. Although I think you're right if you talking about Nvidia hardware only.

Overall: I think it is very clear that any competing VPU manufacturer is more inclined to support MS than NVidia.

[The One]

i would like to see the ati 9700 pro have as much detail as the gfx and the cg look at dawn in this image and look at the deatil it has

[The One]

[PHP]

[The One]

http://www.hothardware.com/reviews/images/gffxshowcase/nymph.jpg
this is great image quality from the gfx i know the benchis suck but i think nvidia has a head start on ati for things to come in the future

[Hanners]

Originally posted by Chalnoth
In fact, if it (Cg) is widely-supported, the performance would likely be higher (on any graphics card) than MS's own HLSL.

Why?

[StealthHawk]

Originally posted by Hanners
Why?

if Cg is highly supported that means other vendors will probably provide optimized compilers for their cards too.

[Hanners]

Originally posted by StealthHawk
if Cg is highly supported that means other vendors will probably provide optimized compilers for their cards too.

I'm pretty sure hell will freeze over before ATi consider adopting Cg, especially when it's functionality is so similar to Microsoft's DirectX 9 HLSL.

[scott123]

I think the biggest problem most folks are having, is the design of the FX. Nvidia hyped this thing so much, and now that it has been released for review, its looking like a major engineering flaw. The heat the FX produces seems so excessive, that you wonder why Nvidia forged foward in the development process when they knew it was going to run like a red-hot frying pan. Nvidia also knew it was not exceeding the performance of the 9700 to any degree, and in some cases, fell behind. For me personally, I smell 3dfx all over this, and it kinda stinks. If it wasn't for the heat, the FX would be getting a lot more favorable reviews.

I think Nvidia had to raise the clock rate far beyond their origional specification, so basically we are looking at an ULTRA x 3 project here. When they make the software underclock during 2D applications, or when the card is supposedly overheating, it makes one wonder if this is a heavily overclocked and overvolted GPU in the first place.

I can't imaging playing something like Nascar 2003 which stresses VGA cards to thier limits, only to see my FPS drop down. Realizing its heat, having to exit the game and allow the card to cool. This sure seems like a nightmare for Nvidia if you ask me. What about people who put this card together like PNY/MSI/BFG who will be providing life-time warranties, I'll bet they aren't feeling too confortabe right now.

My Comp USA pre-order is still there, but as the days go by, Im realizing it may be best to cancel and spend my money elsewhere. I think a lot of other folks are doing the same.

This may sound crazy but I wouldn't be surprised to see Nvidia in the next few weeks scap the entire project, and never release the FX. They know what the reviews and the general public are saying, so they already know its a disaster.

Scott

[Hanners]

Originally posted by scott123
This may sound crazy but I wouldn't be surprised to see Nvidia in the next few weeks scap the entire project, and never release the FX. They know what the reviews and the general public are saying, so they already know its a disaster.

Your right - That is the craziest thing I've ever heard! :D

[Onde Pik]

Originally posted by The Baron
Would you rather be dependent on nVidia or MS? 'Nough said.

Of cource I would, any person in his right mind should when it comes to grahpics. Do u really believe that u can trust Nv over MS to not favor a certain GPU company? ROFL

Originally posted by The Baron
Not if other companies aren't stupid and fail to write Cg compilers if it becomes popular.


Stupidity has nothing to do with it, ATI will have to wait for NV to make every new revision of Cg, do u really think that developers are gonna delay their product to run their development on ATI hardware rather than NV hardware? If u do... maybe studpidity has something to do with it afterall...


Originally posted by The Baron
As Chal said, it's probably faster.

He said if it was widely supported. The same goes for MS HLSL, if THAT is gonna be more widely supported it is probably gonna be faster than Cg. And with all the other drawbacks why the hell should we hope for the biggest support for Cg?



On a general note. It is NOT healthy for the competition to have one company that all the competitors are actually dependant on.