3DMark05 uses PS3...haha

[tranCendenZ]

Have you noticed the amount of people still running GF4s? If you got your head out of your ass, went over to the Valve homepage, and looked at their survey figures, you'd notice that 63% of the people that will be buying HL2 are running GF4s.

That's a skewed survey in favor of ATI as well, since a crapload of people got coupons for HL2 with their ATI cards and subsequently signed up for Steam as a result.

 

[tranCendenZ]

At the time of release the 6800 will be like a geforce 4 ...so you will need to buy a new card...so STFU

No need for inflammatory posts.

Anyway, your analogy is flawed. The ti4200 did not support sm2.0 for 3dmark03, the best it had to offer; but the 6800 supports sm3.0 for 3dmark05, the best it has to offer.

A better analogy will be the x800 will be like a ti4200 while a 6800 will be like a 9700PRO comparatively to 3dmark03; the 6800 not as fast as the current offerings but supporting the same level of shader technology.

 

[^eMpTy^]

lol...XGI will have PS3 before ATi...lol...that's just funny...

and hey, let's not forget that there isn't one single game out that uses PS2.0B...except the FarCry patch which was recalled because it crashed on ATi cards...and for that matter...no games have been announced to be using it either...

 

[rancor]

At the time of release the 6800 will be like a geforce 4 ...so you will need to buy a new card...so STFU

calm down dude. The 6800 gt, ultra are more like like that 9700 from ATi they will last a good 2 years. Games being made right now are made to run on these cards. There won't be any ground breaking speed improvements for gfx cards like we saw this round in the next line.

 

[CrimandEvil]

Have you noticed the amount of people still running GF4s? If you went over to the Valve homepage, and looked at their survey figures, you'd notice that 63% of the people that will be buying HL2 are running GF4s.

The VST sort of looks like Valve is trying to sell ATI cards to them, lol.

 

[rancor]

No need for inflammatory posts.

Anyway, your analogy is flawed. The ti4200 did not support sm2.0 for 3dmark03, the best it had to offer; but the 6800 supports sm3.0 for 3dmark05, the best it has to offer.

A better analogy will be the x800 will be like a ti4200 while a 6800 will be like a 9700PRO comparatively to 3dmark03; the 6800 not as fast as the current offerings but supporting the same level of shader technology.

Opps didn't see your post

 

[Mitolo]

More worthless benchmark BS.

AGREED

 

[joemama]

Why can't Futuremark get some of the game companies (Valve,Id etc..) to just donate some short demos using actual game engines? I'm not one of the people who lives and dies by my 3Dmark score but I wouldn't call the whole thing worthless as a test of videocard capabilities if there were some real world tests thrown in..

 

[tranCendenZ]

Why can't Futuremark get some of the game companies (Valve,Id etc..) to just donate some short demos using actual game engines? I'm not one of the people who lives and dies by my 3Dmark score but I wouldn't call the whole thing worthless as a test of videocard capabilities if there were some real world tests thrown in..

Valve already has their own synthetic benchmark... btw the answer to your question is licensing costs.

 

[burningrave101]

Why can't Futuremark get some of the game companies (Valve,Id etc..) to just donate some short demos using actual game engines? I'm not one of the people who lives and dies by my 3Dmark score but I wouldn't call the whole thing worthless as a test of videocard capabilities if there were some real world tests thrown in..

Have any of you heard of a benchmarking program called Benchemall? If you havn't you should check it out.

http://www.benchemall.com/

 

[LyCoS]

Have you noticed the amount of people still running GF4s? If you went over to the Valve homepage, and looked at their survey figures, you'd notice that 63% of the people that will be buying HL2 are running GF4s..

but then 26% of the total players are using NV17s (geforce 2 and geforce 4mx) that dont even support shaders...

 

[Jonsey]

I think they should stop making these stupid benchmarks and make games...or make a technology game developers can add to thier games to provide easy support for scripted benchmarking...

As it stands now they make cute demos...the numbers are pretty useless...I mean how worthless will PS3 results be? ATi doesn't have it...how can you even compare...

Perfect example of how synthetic benchmarks are useless...IMHO

The crazy thing is that there are lots of people who buy hardware just to "bench" it. Nerd drag racing.

 

[The Batman]

but then 26% of the total players are using NV17s (geforce 2 and geforce 4mx) that dont even support shaders...

Well yes, my point was that not everyone upgrades once every six months. I myself won't be upgrading until the NV50, right around the time the first Unreal 3 based game comes out. As such you'd better damned believe SM3 was an important selling feature. I don't want to get shafted because Nvidia is buying off developers left and right.

 

[Merlin45]

None of the tests in the next 3DMark will actually require SM3.0, but if the user has a graphics card capable of doing SM3.0, it will be used by default. There is actually a new option in the settings, from where the user can choose which shader model will be used (2_0, 2_a, 2_b or 3_0). I think that widens the use of the next 3DMark even more as you can for example compare your graphics cards SM2.0 and SM3.0 performance.

all systems produce the same rendering

from http://www.beyond3d.com/interviews/fm04/index.php?p=2

 

[rancor]

"The next 3DMark will use a completely new 3D engine, which dynamically builds HLSL shaders. The shaders are dynamically built and runtime compiled using the most optimal target for the hardware in use. "

LOL love the way how the guys says, dynamically built and runtime compiled, they do that anyways! its not something special with thier new 3dmark engine

Thats the nice thing about sm 2.0 to 3.0 all shaders are forward, backward compatibale, just performance variences depending on which path you take.

 

[Merlin45]

no, they dynamically compile currently, they don't build the shaders dynamically, they were all hand coded ahead of time, no handcoded shaders in 3dmark next

 

[rancor]

no, they dynamically compile currently, they don't build the shaders dynamically, they were all hand coded ahead of time, no handcoded shaders in 3dmark next

Tell me how are they going to dynamically build shaders? You need a pretty powerful ai to do something like that......

I only know 3 ai's in the world that code c++ let alone asm type languages descently to be even concidered legable code.

I think what they are saying is dynamically constructing shaders for the different paths. Not building them from scratch.

Another possiblity is that they give a set a variable shader to the engine and the engine puts the shaders together. Which is fairly easy to do and cuts down on amount of hand written shaders.

 

[DarkDonut]

i didn't sift through the !!!!!!ism in this thread, but I will say this. It's already rumored that ATi's next chips will have full PS3.0 support and be out Q1 or Q2 of 2005. So the whole PS3.0 debate is squashed as far as i'm concerned.

 

[Merlin45]

(I don't know how this actually works, this is just a guess) they have one master shader, and the engine writes it into the appropriate SM, then compiles it. This seems to make sense and match what they are saying.

 

[rancor]

(I don't know how this actually works, this is just a guess) they have one master shader, and the engine writes it into the appropriate SM, then compiles it. This seems to make sense and match what they are saying.

that is an interesting concept, but I just don't see how it will be as good as optimizing a shader by hand. There's alot of optimizing that can be done. In truth almost anything done with ps 2.0 can be done on ps 1.0 with less of a performance hit. The only real benefits I saw with ps 2.0 was its blending capablities and multi texturing within a shader. But optimizating 2.0 shaders to 1.0 is a very hard thing to do even by hand.

 

[R1ckCa1n]

At the time of release the 6800 will be like a geforce 4 ...so you will need to buy a new card

You need to stop pointing out the obvious!

 

[Merlin45]

that is an interesting concept, but I just don't see how it will be as good as optimizing a shader by hand. There's alot of optimizing that can be done. In truth almost anything done with ps 2.0 can be done on ps 1.0 with less of a performance hit. The only real benefits I saw with ps 2.0 was its blending capablities and multi texturing within a shader. But optimizating 2.0 shaders to 1.0 is a very hard thing to do even by hand.

well, If you read they said that they aren't going to be doing much optimising, also this method guarantees a mathematically equivilant shader from all 4 code paths, which is a big advantage given that that is their goal with the multiple codepaths.

 

[rancor]

well, If you read they said that they aren't going to be doing much optimising, also this method guarantees a mathematically equivilant shader from all 4 code paths, which is a big advantage given that that is their goal with the multiple codepaths.

There never is a mathmatically equivilent in different paths. Doesn't really work that way . Since the API constructs are different for each path, it can never happen, thier code could be mathmatically equivlent, but they have no control of the API.

 

[PadanFain]

First you already have games COMING OUT that announced they WILL BE using 3.0

so save that it won't be used... that is a bold faced lie

and ATI might have been better throwing futuremark some of that 6 million they gave to valve

 

[Merlin45]

well, given that all these paths are subsets or supersets of one another (it could be argued that all the paths are SM 3.0 subsets) producing mathematically equivilent results in the reference renderer is very possible, the problem comes with producing the same results in the various hardware available, as that has variables outside of the control of futuremark. you could produce a 2 pass SM2.0 shader that would be mathematically equivilent to a 1 pass SM 3.0 shader. while in hardware it would loose some precision from being multipassed, in reference it would be identical in output. (it must be possilbe as the far cry devs have said that their SM 2.0, 2.0b, and 3.0 per pixel lighting paths are all mathematically equivilant)

 

[rancor]

exactly, if the same shader is that is hand written is used for SM 2.0, ps 2.0b and sm 3.0 it will be mathmatically equivlent as possible because of that API factor.

Far Cry was a good demonstration of that specially with thier light shader which was the same shader used for all paths but there was a varience in gloss levels from one path to another.

 

[LyCoS]

Well yes, my point was that not everyone upgrades once every six months. I myself won't be upgrading until the NV50, right around the time the first Unreal 3 based game comes out. As such you'd better damned believe SM3 was an important selling feature. I don't want to get shafted because Nvidia is buying off developers left and right.

so basically, i who upgrade every time a graphics card comes out (i have like : 1 9700 *fried*, 9700pro*fried*, 2 9800pro *one fried*, 2 9600XT , 1 geforce 6800gt and 2 X800XT PEs)) am not concernet

one thing that is forgotten :

ATIs shader calculation is like 8x faster then NVIDIAs. so if nvidia goes twice a fast with SM 3.0, ATI is still 4x faster for shaders.. SM 3.0 aparently doesnt bring any visual features, just speed... get my point?

 

[SnakEyez187]

and ATI might have been better throwing futuremark some of that 6 million they gave to valve

Oh how impartial, Nvidia never does these such things, good comment sir

 

[rancor]

so basically, i who upgrade every time a graphics card comes out (i have like : 1 9700 *fried*, 9700pro*fried*, 2 9800pro *one fried*, 2 9600XT , 1 geforce 6800gt and 2 X800XT PEs)) am not concernet

one thing that is forgotten :

ATIs shader calculation is like 8x faster then NVIDIAs. so if nvidia goes twice a fast with SM 3.0, ATI is still 4x faster for shaders.. SM 3.0 aparently doesnt bring any visual features, just speed... get my point?

lol try again take a look around the shadermark tests for sm 2.0 with the gf 6 vs x800 line

depends on the type of shader vertex or pixel (ati is better at vertex, nV is better at pixel). At the end its very close.

 

[rancor]

Oh how impartial, Nvidia never does these such things, good comment sir

Well for Valve to take that offer is kinda pathetic...... A game company in that posistion doesn't need to take it. They would have gotten that money anyways.

 

[The Batman]

so basically, i who upgrade every time a graphics card comes out (i have like : 1 9700 *fried*, 9700pro*fried*, 2 9800pro *one fried*, 2 9600XT , 1 geforce 6800gt and 2 X800XT PEs)) am not concernet

one thing that is forgotten :

ATIs shader calculation is like 8x faster then NVIDIAs. so if nvidia goes twice a fast with SM 3.0, ATI is still 4x faster for shaders.. SM 3.0 aparently doesnt bring any visual features, just speed... get my point?

Links. Last time I checked the benchs the NV40 and X800 were pretty close in shader speeds. Get my point?

Also, about the other topic, Valve isn't as rich/powerful as people make them out to be. They've had one successfull game [which wasn't entirely their own engine in the first place] and haven't really marketed their engines yet [we got Vampire but that's about it so far, though I do here rumours of another Source RPG]. In other words Valve is no id, if I were valve I'd jump at 6 million. I don't blame them for it, I blame them for being incompetant with their schedules.

 

[LyCoS]

First you already have games COMING OUT that announced they WILL BE using 3.0

so save that it won't be used... that is a bold faced lie

and ATI might have been better throwing futuremark some of that 6 million they gave to valve

they did and so did nvidia. To even have you're graphics card working with 3D mark you have to pay a fortune (hence the *impartiality* of the 3D mak benchmark...). I think hardocp.com wrote an exellent article on that, how you pay like $200 000 to influence the bench, and a few thousand to know what is happening.... THAT is why 3D mark isnt very credible...

 

[SnakEyez187]

Well for Valve to take that offer is kinda pathetic...... A game company in that posistion doesn't need to take it. They would have gotten that money anyways.

Every game company does it, if you're going to bundle a game with any of your cards, put up any advertising with the game's name or publishers name on your website or on any boxes, you are going to have to pay them to do so

 

[rancor]

Every game company does it, if you're going to bundle a game with any of your cards, put up any advertising with the game's name or publishers name on your website or on any boxes, you are going to have to pay them to do so

they don't get that kind of cash though. Usually its just some development software, graphics cards what not. Not hard cash.

 

[^eMpTy^]

Oh how impartial, Nvidia never does these such things, good comment sir

hahaha...YOU accused someone of being biased...that's laughable...

and there was no implication that throwing around money is a bad thing...you thought of that all by yourself...all he said was maybe they should have tossed some of that valve money in some other places, since (according to the latest benchmarks) they didn't really get thier money's worth...hahahaha

 

[PadanFain]

they did and so did nvidia. To even have you're graphics card working with 3D mark you have to pay a fortune (hence the *impartiality* of the 3D mak benchmark...). I think hardocp.com wrote an exellent article on that, how you pay like $200 000 to influence the bench, and a few thousand to know what is happening.... THAT is why 3D mark isnt very credible...

that was sarcasm...

at the conduct of paying a game creator 6 million....

I don't even run 3dmark so that should tell you how I feel about it. 3.0 is still going to be used in upcoming games as about 5 have already announced they WILL BE using it sp

I don't need futuremark to tell me that

 

[R1ckCa1n]

So since yet another topic has turned to my SM3.0 does this and yours doesn't, how well will the NV40 perform in the up-and-coming SM3.0 games ? How will SM3.0 increase IQ? Since it's soooooo good, I'd like to see some real data.

 

[PadanFain]

So since yet another topic has turned to my SM3.0 does this and yours doesn't, how well will the NV40 perform in the up-and-coming SM3.0 games ? How will SM3.0 increase IQ? Since it's soooooo good, I'd like to see some real data.

so would I but only a idiot would arguing NOT having it is a good thing

I don't think game devs would use it or announce to using it in upcoming games if it sucked that bad

 

[Brent_Justice]

i hope games move to SM 3.0 pretty fast, i want 2005 to be the year of SM 3.0

cause its gonna be a long time till SM 4.0

2006-2007 till Longhorn comes out

not to mention games developed to use SM 4.0, could be 2007-2008

 

[burningrave101]

that was sarcasm...

at the conduct of paying a game creator 6 million....

I don't even run 3dmark so that should tell you how I feel about it. 3.0 is still going to be used in upcoming games as about 5 have already announced they WILL BE using it sp

I don't need futuremark to tell me that

http://www.nvnews.net/vbulletin/showthread.php?t=30736

There has been at least 13 announce SM 3.0 support so far. There may be more now.