Introduction to 3DMark2003

I think DT is making two unstated assumptions here:

1. Because Stalker is using Cg, it will not have any PS 1.4 shaders;

2. Not using PS 1.4 shaders will make the game relatively unoptimized (by which I mean not as well optimized as for other platforms) on R200, RV250, and probably R300 hardware.

Assumption 1 seems safe, unless Nvidia has added PS 1.4 support to Cg. (Have they? I'm ignorant on this point.) But only to the extent that the programmers rely exclusively on Cg; they can always add their own PS 1.4 code afterwards.

Assumption 2 seems the bigger leap, as it assumes the game uses effects which require multiple passes with PS 1.1 but can be single-passed using PS 1.4. Right now I think it's safe to say we have no idea whether this is the case.

Bottom line: the heavy Nvidia-tilted marketing for this game appears unseemly, but very likely it is just a case of an unknown developer trying to drum up press/buzz for their game (and succeeding). Past experience with these sorts of games is mixed: some turn out to be significantly better targeted at Nvidia hardware, and some turn out neutral.

Why don't we wait and see?

 

The discussion about boycotting vendor specific titles is not quite hitting the mark.

It's perfectly okay (and one might argue that even preferable) for game developers to optimize their hardware to any specific platform the like. There's nothing wrong with that. Period. Game developer typically does not need to worry about things that are related to performance measurement. Put out a good game, get additional visibility and if it's used as a benchmark, maybe even make an extra hundred thou' if a major vendor wants to pay for doing the additional optimizations. Everybody wins, right?

The real issue that comes into play is when the same vendor then promotes that optimized game as the 'major' 3D benchmark. Typically there's not technical documentation, and with exception of JC/DoomIII it's practically impossible even to dig up what optimizations have been done for that title. I for one have not seen a single technical document that would exlain the inner workings of any game benchmark (again JC being an exception to the rule).

Imagine if there would not be _any_ neutral benchmark code to serve as a reference. You'd have one of the companies arguing that Commance4 is the one and only answer, while the other would argue that Serious Sam 'does things right'. Neither of the those two have actually been real triple A titles, yet both are being used as 'benchmark showcases', which is just plain downright ridiculous when you think about it.

This kind of 'divide and conquer' tactics would only lead into a situation where the biggest marketing budget dictates what is 'perceived as good'. Then you'd be able to sell inadequate hardware just with bigger and better marketing slogans.

Games are great as benchmarks, but with the exception of JC, very few developers do public disclosures of what's gone in them and why.

 

Nice post, guest!

 

well now

take a look at this screen itrs using Nvidas CG CinFX engine.

http://www.pixelrage.ro/images/large/stalker_1.jpg

 

I've seen better done on ATI hardware...

 

no

Brite i dpont think so can ou show me a image of somthing done for ATI

 

I can... and it was done 8 months ago on a 9700

 

Don't feed the trolls and they'll look elsewhere for sustenance.

 

ommg so lame

those images of Doom 3 were on a 9700 but infact ther is no opt. for it at all in the game i took some images to on a GeForce 4 Ti 4400

 

ok i dont know how to insert a image so oh well just add me to your friends list in yahoo
the_matrix_overload

 

oh yea nice cut scene shots and of images fousn oj n the net
i got some from my system not on the net

 

I prefer the stalker pic, more realistic IMO.

 

look here

http://www.digit-life.com/articles2/gffx/gffx-ref-p9.html


in that link at the bottem is Doom3 test agianst both ATI 9700 and GFFX and as ya can see the GFFX is about 3 times faster for Doom 3

 

Re: look here

BULLSHIT!

excuse the term.. but it really is...
This Alpha is sooooo oooooold.

We have infos from JC directly who "should" have the best information available on the planet... and who is not using a 1 year old engine demo.....

 

There is no such thing as a Doom3 test, just some leaked "not-even-alpha" version that was used to demo the engine at last year's E3.

If you read Carmack's latest plan, the GeForceFX ULTRA (which much is more expensive than the 9700 pro...if you can buy an Ultra at all) is just SLIGHTLY faster than the 9700 Pro.

At the same time, the 9700 Pro uses MORE ADVANCED features than the FX, because if you use those features on the FX (the ARB2 rendering path), the performance of the FX cuts in HALF.

 

well i reead hois last plan and it great but the GFFX is optamized for Doom 3 ther is no optamization for 9700 and if ya look in that site on that test tho its alpha version witch is not optamized yet to run on a PC that well i think its a good test at the power the GFFX really has. and if you read his plan you would also see wher he says.

I am using an NV30 in my primary work system now, largely so I can test more
of the rendering paths on one system, and because I feel Nvidia still has
somewhat better driver quality the NV30 runs fragment programs much slower, but it has a huge maximum
instruction count. I have bumped into program limits on the R300 already

 

oh and

oh and hers a list of what i see in the Doom 3 config file for the alpha

ythis is closeot the ending of it


seta r_shadowMapBias "0"
seta r_specular "1"
seta r_debugLineWidth "1"
seta r_debugLineDepthTest "0"
seta r_forceLoadImages "0"
seta r_useStandardGL "0"
seta r_useParhelia "0"
seta r_useNV20 "1"
seta r_useNV30 "0"
seta r_useGL2 "0"
seta r_lightScale_ "2"
seta r_shadows "1"
seta r_gammaScale "1"
seta r_gammaBias "2"
seta r_stepGamma5 "16"
seta r_stepGamma4 "32"
seta r_stepGamma3 "196"
seta r_stepGamma2 "64"
seta r_stepGamma1 "128"
seta r_brightness "1"
seta r_gamma "1"
seta r_swapInterval "1"
seta r_customHeight "486"
seta r_customWidth "720"
seta r_multiSamples "0"
seta r_stereo "0"
seta r_depthbits "24"
seta r_stencilbits "8"
seta r_colorbits "32"
seta r_ext_vertex_array_range "1"
seta r_ext_texture_env_combine "1"
seta r_ext_texture_edge_clamp "1"
seta r_ext_texture_filter_anisotropic "0"
seta r_ext_compress_textures "0"
seta com_drawfps "1"
seta r_mode "4"
seta r_fullscreen "1"

 

Re: look here

yeah, look again :

 

poo

heres my email uarehere18@hotmail.com you can contact me on msn messanger or yahoo messanger for info on the stuff i posted and all send Doom3 to you if you want it to see for yourself my yahoo name in messanger is
the_matrix_overload

 

My email reply...


Hi,

In fact, the graphics engine is developed on Radeon9700
We demand special support from FX driver guys, 'cause in pure/standart
DX9 it's impossible so run the engine on FX, at the moment.
Yes, FX will be slightly faster, but in the margin of several
percentages...

So, don't worry, your card will be excelent performer in
S.T.A.L.K.E.R.

--
Best regards,
Oles V. Shishkovtsov
GSC-Game World
oles@gsc-game.kiev.ua


Monday, February 17, 2003, 7:05:06 PM, you wrote:

OVJ> This is a forwarded message
OVJ> From: Anton Bolshakov <anton@gsc-game.kiev.ua>
OVJ> To: "Oleg V . Javorsky" <oleg@gsc-game.kiev.ua> (GSC Game World)
OVJ> Date: Monday, February 17, 2003, 7:00:11 PM
OVJ> Subject: : Stalker Support


Doh sorry about that...screen shots are taken on a 9700.