Visual differences between Pixelshader 2.0(b) and 3.0...

[razor1]

I think your really as streching this. Yes some will run slower, but to say that every effect done in SM3.0 will be too slow to run in SM2/2.0b is not correct as your post is making it sound...

not really tell me what were the first cards that could do full screen of normal mapping? GF3, but starting with dx 6 maybe 7 been too long, there were normal mapping demo's. So does that mean the GF 2's could do the same IQ level as the GF3 sure, but is it playable. No. You take self shadowing parrallex bump mapping and try that on a sm 2.0b card, it would be crushed. Do it on any sm 3.0 card, well almost any, would say 6800 gt and up, it won't have much issue with it at all. The thing is for a developer to add these features they have to look at who has what cards. I'm not going to spend time on optimizing a shader or making another path for less then 20% of the people that are going to be playing the game. If they want to play the game they better upgrade thier cards. Since the only 3 cards that would have had the processing power to the effects anyways, x8xx XT and XT PE, and xl (they possibly can't do these shaders that we are looking at in these games too at a reasonable speed, so the developers left them out for a reason), not a whole lot of people have these cards since they are top end, would you speed 4 -5 months as a game company to make these. Thats about a 500k you would be spending for salaries, maybe more. Since sm 2.0b is an interm shader model, it wasn't adopted well, unfortunate but an oversite by ATi.

 

[Netrat33]

not really tell me what were the first cards that could do full screen of normal mapping? GF3, but starting with dx 6 maybe 7 been too long, there were normal mapping demo's. So does that mean the GF 2's could do the same IQ level as the GF3 sure, but is it playable. No. You take self shadowing parrallex bump mapping and try that on a sm 2.0b card, it would be crushed. Do it on any sm 3.0 card, well almost any, would say 6800 gt and up, it won't have much issue with it at all.

What are you talking about? SPCT doesn't crush x800 cards when it's used. FEAR puts last year generation practically in the same both and it uses SM2.

The thing is for a developer to add these features they have to look at who has what cards. I'm not going to spend time on optimizing a shader or making another path for less then 20% of the people that are going to be playing the game. If they want to play the game they better upgrade thier cards. Since the only 3 cards that would have had the processing power to the effects anyways, x8xx XT and XT PE, and xl (they possibly can't do these shaders that we are looking at in these games too at a reasonable speed, so the developers left them out for a reason), not a whole lot of people have these cards since they are top end, would you speed 4 -5 months as a game company to make these. Thats about a 500k you would be spending for salaries, maybe more. Since sm 2.0b is an interm shader model, it wasn't adopted well, unfortunate but an oversite by ATi.

x700, x800gto , x800gto2 (you get the idea) and so on and so on all use sm2.b which are cheap cards and sold very well. And yes you ALREADY see game developers using sb2b which you see with these 2 games and only one of those having an image quality difference. SPCT, farcry, B&W2, AoEIII, Unreal 3 reported to use it, and I feel i'm missing 3 more games, but you see that it's being used. About the same rate as sm3 games have been.

 

[Jbirney]

not really tell me what were the first cards that could do full screen of normal mapping? GF3, but starting with dx 6 maybe 7 been too long, there were normal mapping demo's. So does that mean the GF 2's could do the same IQ level as the GF3 sure, but is it playable. No. You take self shadowing parrallex bump mapping and try that on a sm 2.0b card, it would be crushed. Do it on any sm 3.0 card, well almost any, would say 6800 gt and up, it won't have much issue with it at all.

Again I am sure you can find a few cases where an effects done in PS3.0 is "un-playable" on a card that only supports PS2.0 But those so far have not show up in an games yet.

We also have seen in the past where a more advacned PS card was still slower. The Adv PS test of 3Dmark2001 still had the GF4 faster than the 8500 even though the 8500s could do the shaders in one pass vrs the multi-pass GF4 took. Farcy again showed us simular things that the 800s cards were still faster even though the 6800s only needed one pass at the shaders.

My point was I felt your made to broad of a generalization saying that every thing would be too slow to run and the fact is thats not the case. Plenty of other things factor in to the over all proformance numbers and we have lots of data to show that in games so far having to render things one (or few) pass does not mean its the fastest card in that game...

The thing is for a developer to add these features they have to look at who has what cards. I'm not going to spend time on optimizing a shader or making another path for less then 20% of the people that are going to be playing the game. If they want to play the game they better upgrade thier cards.Since the only 3 cards that would have had the processing power to the effects anyways, x8xx XT and XT PE, and xl (they possibly can't do these shaders that we are looking at in these games too at a reasonable speed, so the developers left them out for a reason), not a whole lot of people have these cards since they are top end, would you speed 4 -5 months as a game company to make these. Thats about a 500k you would be spending for salaries, maybe more. Since sm 2.0b is an interm shader model, it wasn't adopted well, unfortunate but an oversite by ATi.

Funny this is why valve dropped their DX9 path for the FX cards and IIRC I did not remember you backing that choice back then as your just saying the same thing that they said

Also as more of techincallity any PS3.0 card can run all the lower shader models, thus a PS3.0 can run PS2.0b and that makes the install base of PS2.0b larger than PS3.0 cards

 

[eno-on]

Seriously, razor. Heres the question. There needs to be a one word answer, being yes or no.

Is there an IQ difference between 2.0b and 3.0 if they are programmed for?

If you say anything other than 'no', you'd be lying. There are no qualifying remarks involved in this. The answer, in regards to the statement made by Terra is NO. THERE ARE NO IQ DIFFERENCES BETWEEN 2.0b and 3.0. Period. To argue differently because the moon is blue on monday or game X does not use 2.0b to it's potential is being dishonest, erroneous and misleading. So FFS knock it off and admit there is no goddamned difference in IQ. nVidia will still love you if you admit it, I promise.

 

[Trimlock]

nVidia will still love you if you admit it, I promise.

hahahaha

 

[Jbirney]

nVidia will still love you if you admit it, I promise.

LMAO!!!!

 

[razor1]

Again I am sure you can find a few cases where an effects done in PS3.0 is "un-playable" on a card that only supports PS2.0 But those so far have not show up in an games yet.

We also have seen in the past where a more advacned PS card was still slower. The Adv PS test of 3Dmark2001 still had the GF4 faster than the 8500 even though the 8500s could do the shaders in one pass vrs the multi-pass GF4 took. Farcy again showed us simular things that the 800s cards were still faster even though the 6800s only needed one pass at the shaders.

My point was I felt your made to broad of a generalization saying that every thing would be too slow to run and the fact is thats not the case. Plenty of other things factor in to the over all proformance numbers and we have lots of data to show that in games so far having to render things one (or few) pass does not mean its the fastest card in that game

...

3dmark 01 yeah, the GF 4 was faster, thats because it had the computational power to over come the features deficient. Thats what I've been saying wasn't it? This is an old arguement that goes both ways.

A few now may later. Thats the way things go. People are going to be programming effects that can be played on a given card unless the card is there to begin with and even then it takes a year or two to make those changes from the release of such a card.

You are reading into it wrong. The radeons do better in far cry not do to pixel shader performance. Its do to vertex shader performance. Don't even get into this you know the radeons are weak comparied to the gf6 and 7 pixel shader performance other then one or two tests where they are superior out of the many.

Funny this is why valve dropped their DX9 path for the FX cards and IIRC I did not remember you backing that choice back then as your just saying the same thing that they said

Also as more of techincallity any PS3.0 card can run all the lower shader models, thus a PS3.0 can run PS2.0b and that makes the install base of PS2.0b larger than PS3.0 cards

Well look at it this way a guy that made a mod was able to get water to look just fine on the gf fx's with very little speed loss. You think that was too hard for the developer to do that? Mod makers have very little access to engine code, so if it was that easy to do why didn't Valve do it? . This was not a new path or a whole new shader for that matter, just one shader that was tweaked. I call that laziness that Valve who spent 5 years in making a game that should have taken them 2, and they couldn't tweak a shader that would have taken them a day or two possibly even less? You are taking about some of the best programmers in the world here, not a modder.

Who said that PS 3.0 cards can't run older paths. But do people want to support lagacy paths, that not many people would have use for?

 

[razor1]

Seriously, razor. Heres the question. There needs to be a one word answer, being yes or no.

Is there an IQ difference between 2.0b and 3.0 if they are programmed for?

If you say anything other than 'no', you'd be lying. There are no qualifying remarks involved in this. The answer, in regards to the statement made by Terra is NO. THERE ARE NO IQ DIFFERENCES BETWEEN 2.0b and 3.0. Period. To argue differently because the moon is blue on monday or game X does not use 2.0b to it's potential is being dishonest, erroneous and misleading.

Your argument is pointless and frivilous. I would say NO they can't period. As I said before. And gave you the reason. And told you this has been said many times over and over again. Go ahead and try HDR on a 9700 or even a 9800 in game, I'm sure you are this interesting take Far Cry patch 1.3 or higher and mod it to get sm 2.0 to work with the sm 2.0b HDR . Yeah it will look great but it isn't playable. I would say NO they can't do it. You still haven't given me any good examples of effects that could have been done with very little speed hit so the game will still be playable on sm 2.0b cards. Since there isn't any . Base you arguement on facts, not theories. In truth I know I could make parrallex bump mapping work very fast almost as fast as normal mapping on sm 1.4 hardware. But thats just it how far the programmers want to take it. But would programmers want to spend a week or two doing the optimizations for one shader for a set number people who still have those cards? Not really.

You can in theory say a GF 1 can do the same IQ level as a ATi x1800xt, just will go .5 fps. Oh yeah a Pentium 667 can do raytracing too. Does that mean its equal to the raytracing of the x1800xt or the GF7's? Or how about true radiosity. Yeah it can be rendered through GPU's its been done for ages now, does that mean in 5 years when graphics cards can do it at playable speeds that GPU's should be programmed for to do radiosity, just because it can be there?

So FFS knock it off and admit there is no goddamned difference in IQ. nVidia will still love you if you admit it, I promise.

Why don't you admit that you have no fact to base your opionon on. Oh yeah and take your nose out of ATi's butt

 

[Digital Viper-X-]

Your argument is pointless and frivilous. I would say NO they can't period. As I said before. And gave you the reason. And told you this has been said many times over and over again. Go ahead and try HDR on a 9700 or even a 9800 in game, I'm sure you are this interesting take Far Cry patch 1.3 or higher and mod it to get sm 2.0 to work with the sm 2.0b HDR . Yeah it will look great but it isn't playable. I would say NO they can't do it. You still haven't given me any good examples of effects that could have been done with very little speed hit so the game will still be playable on sm 2.0b cards. Since there isn't any . Base you arguement on facts, not theories. In truth I know I could make parrallex bump mapping work very fast almost as fast as normal mapping on sm 1.4 hardware. But thats just it how far the programmers want to take it. But would programmers want to spend a week or two doing the optimizations for one shader for a set number people who still have those cards? Not really.

You can in theory say a GF 1 can do the same IQ level as a ATi x1800xt, just will go .5 fps. Oh yeah a Pentium 667 can do raytracing too. Does that mean its equal to the raytracing of the x1800xt or the GF7's? Or how about true radiosity. Yeah it can be rendered through GPU's its been done for ages now, does that mean in 5 years when graphics cards can do it at playable speeds that GPU's should be programmed for to do radiosity, just because it can be there?



Why don't you admit that you have no fact to base your opionon on. Oh yeah and take your nose out of ATi's butt

I Think ur taking things out of context, what he means is CURRENT cards, X800 can do what a 6800 Can do, without a big performance hit, stop being jackasses!

 

[eno-on]

Your argument is pointless and frivilous. I would say NO they can't period. As I said before. And gave you the reason. And told you this has been said many times over and over again. Go ahead and try HDR on a 9700 or even a 9800 in game, I'm sure you are this interesting take Far Cry patch 1.3 or higher and mod it to get sm 2.0 to work with the sm 2.0b HDR . Yeah it will look great but it isn't playable. I would say NO they can't do it. You still haven't given me any good examples of effects that could have been done with very little speed hit so the game will still be playable on sm 2.0b cards. Since there isn't any . Base you arguement on facts, not theories. In truth I know I could make parrallex bump mapping work very fast almost as fast as normal mapping on sm 1.4 hardware. But thats just it how far the programmers want to take it. But would programmers want to spend a week or two doing the optimizations for a set number people who still have those cards? Not really.

You can in theory say a GF 1 can do the same IQ level as a ATi x1800xt, just will go .5 fps. Oh yeah a Pentium 667 can do raytracing too. Does that mean its equal to the raytracing of the x1800xt or the GF7's?



Why don't you admit that you have no fact to base your opionon on. Oh yeah and take your nose out of ATi's butt

You would say no because you don't make any sense and are incapable of realising that two plus two does not equal twenty-two.
As was shown on previous hardware configurations with games that came out then, the sm 2.0b did not show any real performance differences from 3.0 when programmed for. Of course that will change with newer titles.

This is NOT MY OPINION. Factually, 2.0b, 2.0, 1.4, etc, are CAPABLE OF THE SAME IQ, just at (sometimes drastically reduced) performance.
I never said that 2.0b would be as fast 3.0 doing the same IQ. I said only that 2.0b was capable of the same IQ, so don't put words in my mouth. This fact is true, that there are no visual differences between what 2.0b adn 3.0 can do. At all. Face it.

In truth I know I could make parrallex bump mapping work very fast almost as fast as normal mapping on sm 1.4 hardware. But thats just it how far the programmers want to take it

Thanks for saying exactly what Ive been saying. I dunno why you want to keep arguing the issue.

The difference in 2.0b and 3.0 on the AOE screenshots IS NOT because 2.0b CAN'T DO IT, it's because no one PROGRAMMED 2.0b to DO IT in AOE.

If you can't agree to the statement above (though you did in the quote above anyway, more or less), then sorry man, we can't have this conversation.

I have agreed with you, numerous times, that coding those effects in 2.0b may drastically reduce the performance of the game. Possibly to where it is unplayable.
That doesn't change the FACT that 2.0b is CAPABLE of the same IQ.
As far as having my nose up ATis butt, who really gives a fuck? For gods sake, they have sm 3.0 parts out now, so that's really irrelevant. I buy whats good and what I can afford.

 

[razor1]

You would say no because you don't make any sense and are incapable of realising that two plus two does not equal twenty-two.
As was shown on previous hardware configurations with games that came out then, the sm 2.0b did not show any real performance differences from 3.0 when programmed for. Of course that will change with newer titles.

This is NOT MY OPINION. Factually, 2.0b, 2.0, 1.4, etc, are CAPABLE OF THE SAME IQ, just at (sometimes drastically reduced) performance.
I never said that 2.0b would be as fast 3.0 doing the same IQ. I said only that 2.0b was capable of the same IQ, so don't put words in my mouth. This fact is true, that there are no visual differences between what 2.0b adn 3.0 can do. At all. Face it.

Thanks for saying exactly what Ive been saying. I dunno why you want to keep arguing the issue.

The difference in 2.0b and 3.0 on the AOE screenshots IS NOT because 2.0b CAN'T DO IT, it's because no one PROGRAMMED 2.0b to DO IT in AOE.

If you can't agree to the statement above (though you did in the quote above anyway, more or less), then sorry man, we can't have this conversation.

I have agreed with you, numerous times, that coding those effects in 2.0b may drastically reduce the performance of the game. Possibly to where it is unplayable.
That doesn't change the FACT that 2.0b is CAPABLE of the same IQ.
As far as having my nose up ATis butt, who really gives a fuck? For gods sake, they have sm 3.0 parts out now, so that's really irrelevant. I buy whats good and what I can afford.

It all depends on what terms we look at it. I look at it from the speed side where something is capable and you look at from pure capability, thats the beauty of silicon chips, you can make them function the way you wish for the most part, so in my eyes you can't look at capability that way you have a goal that goal is play it in real time, so that is where you have to analyze from. This is what I said before. So yes I somewhat agree with you also, but you going around telling people are trolling, when they do have a point, and you not giving any good reasons behind your opnion, Terra gave his, he has the games to thank for that, you didn't. Then you go about saying I have some sort of link with nV. I like your attitude. Very condicending and negetive. Thank you for admitting that you your nose is somewhere where it shouldnt' be.

You and netrat 33 did that, there was no excuse. Give me a viable explination is what I asked, you can't give it.

 

[eno-on]

It all depends on what terms we look at it. I look at it from the speed side where something is capable and you look at from pure capability, thats the beauty of silicon chips, you can make then function them way you wish for the most part, so in my eyes you can't look at capability that way you have a goal that goal is play it in real time, so that is where you have to analyze from. This is what I said before. So yes I somewhat agree with you also, but you going around telling people are trolling, when they do have a point, and you not giving any good reasons behind your opnion, Terra gave his, he has the games to thank for that, you didn't. Then you go about saying I have some sort of link with nV. I like your attitude. Very condicending and negetive. Thank you for admitting that you your nose is somewhere where it shouldnt' be.

You and netrat 33 did that, there was no excuse. Give me a viable explination is what I asked, you can't give it.

I didn't call anyone a troll.
All I'm saying is that the differences between the screenshots in AOE was not because 2.0b was incapable of the effects, only that programmers did not include them for the 2.0b path in that game :\
That was my only point, and it stands as factual.

/edit : btw, I guess I wasn't clear. I do not have my nose up ATi's butt. I need a new card, but haven't decided what yet, but everything is pointing towards a 7800gt for me.
Also, I wa snot implying that you have ties to nV, only that you are somewhat pro-nV, which is evidenced from almost all of your previous posts.

 

[razor1]

I didn't call anyone a troll.
All I'm saying is that the differences between the screenshots in AOE was not because 2.0b was incapable of the effects, only that programmers did not include them for the 2.0b path in that game :\
That was my only point, and it stands as factual.

/edit : btw, I guess I wasn't clear. I do not have my nose up ATi's butt. I need a new card, but haven't decided what yet, but everything is pointing towards a 7800gt for me.
Also, I wa snot implying that you have ties to nV, only that you are somewhat pro-nV, which is evidenced from almost all of your previous posts.

Oh sorry got you and fallguy mixed up. I'm sorry for saying that you had your nose up ATi's butt too, just that I wasn't expecting that from you for all the arguements we had have been fairly civil.

Well thats the problem, 3dsmax or maya can do those effects too without a graphics card. Thats the flaw in the arguement. And thats where I was coming from. I will agree with that sm 2.0b and 3.0 can do the same effects. But not for in game purposes on last years cards as newer games come out this will be more noticable.

I'm pro nV for the time being, but I do see ATi has a great architecture that they can ride on for a while. Its going to be easier for ATi to catch up on pixel shader performance then nV catching up to ATi on thier memory control which will be quite a big change for them. I would say wait a couple of months on your graphics card if you can, something might be coming out soon .

 

[Chris_Morley]

I think that the correct way to look at this, and the one that won't bruise any egos is this carefully constructed sentence:

There is a visual quality difference between 2.0b and 3.0 because to do the same effects in SM 2.0b, while entirely possible, is typically crippling to the generation of cards that are limited to 2.0b; thus the effects are not applied but to cards which are capable of SM 3.0 - which would deliver the same visuals at an acceptable frame rate.

Everybody happy?

 

[eno-on]

Look, what I'm saying is really simple.
Obviously the AOE screenshots look different.
3.0 looks better.
But it's not because of the limitations of sm2.0b that they look better, it's because of the programming in AOE did not program those effects for the 2.0b.
Terra is implying that the screenshots look better due to limitations in sm2.0b. Which is flat out wrong, and misleading.
And it all boils down to the fact that it's a totally moot point, as both ATi and nV have 3.0 hardware out. We wont know if there was a major performance hit by enabling those effects in 2.0b because AOE doesn't give us that option.

 

[eno-on]

I think that the correct way to look at this, and the one that won't bruise any egos is this carefully constructed sentence:

There is a visual quality difference between 2.0b and 3.0 because to do the same effects in SM 2.0b, while entirely possible, is crippling to the generation of cards that are limited to 2.0b; thus the effects are not applied but to cards which are capable of SM 3.0 - which would deliver the same visuals at an acceptable frame rate.

Everybody happy?

Kinda. Though not true in every case. Look at the farcy patch. x800 series did fine under 2.0b while looking identical to the sm3.0 of the 6x00 series.

 

[Chris_Morley]

Kinda. Though not true in every case. Look at the farcy patch. x800 series did fine under 2.0b while looking identical to the sm3.0 of the 6x00 series.

Revised.

 

[razor1]

I think that the correct way to look at this, and the one that won't bruise any egos is this carefully constructed sentence:

There is a visual quality difference between 2.0b and 3.0 because to do the same effects in SM 2.0b, while entirely possible, is typically crippling to the generation of cards that are limited to 2.0b; thus the effects are not applied but to cards which are capable of SM 3.0 - which would deliver the same visuals at an acceptable frame rate.

Everybody happy?

Very well put.

 

[Netrat33]

It all depends on what terms we look at it. I look at it from the speed side where something is capable and you look at from pure capability, thats the beauty of silicon chips, you can make them function the way you wish for the most part, so in my eyes you can't look at capability that way you have a goal that goal is play it in real time, so that is where you have to analyze from. This is what I said before. So yes I somewhat agree with you also, but you going around telling people are trolling, when they do have a point, and you not giving any good reasons behind your opnion, Terra gave his, he has the games to thank for that, you didn't. Then you go about saying I have some sort of link with nV. I like your attitude. Very condicending and negetive. Thank you for admitting that you your nose is somewhere where it shouldnt' be.

You and netrat 33 did that, there was no excuse. Give me a viable explination is what I asked, you can't give it.

Did you even look at the benchmarks on the tom's hardware...notice B&W2 for sm2.0b and 3....not really much of a difference. 1 frame.

Sm3 and sm2.b on AoE3, X800 cards beating 6800, meaning there's more room for improvement if so desired. So ya..it's a moot point.

 

[Netrat33]

I think that the correct way to look at this, and the one that won't bruise any egos is this carefully constructed sentence:

There is a visual quality difference between 2.0b and 3.0 because to do the same effects in SM 2.0b, while entirely possible, is typically crippling to the generation of cards that are limited to 2.0b; thus the effects are not applied but to cards which are capable of SM 3.0 - which would deliver the same visuals at an acceptable frame rate.

Everybody happy?

Show me a game where it was CRIPPLING it.

 

[Netrat33]

Look, what I'm saying is really simple.
Obviously the AOE screenshots look different.
3.0 looks better.
But it's not because of the limitations of sm2.0b that they look better, it's because of the programming in AOE did not program those effects for the 2.0b.
Terra is implying that the screenshots look better due to limitations in sm2.0b. Which is flat out wrong, and misleading.And it all boils down to the fact that it's a totally moot point, as both ATi and nV have 3.0 hardware out. We wont know if there was a major performance hit by enabling those effects in 2.0b because AOE doesn't give us that option.

ding ding ding.

 

[razor1]

Did you even look at the benchmarks on the tom's hardware...notice B&W2 for sm2.0b and 3....not really much of a difference. 1 frame.

Sm3 and sm2.b on AoE3, X800 cards beating 6800, meaning there's more room for improvement if so desired. So ya..it's a moot point.

If I'm not mistaken AOE3 does HDR or another post render effect, pertty sure it was blur for the 6800's under sm 3.0, 6800's take a large hit from this so will the x800's if they did it too. Also they didn't test AOE 3 in that review for the x800's only on the 6800's did they test . . Yes keep rolling your eyes since that is all you can do.

Black and White can't compare those benchmarks yes very little difference but different paths, the gf's are only back by 1 fps and a bit more on higher settings but they are doing alot more. If that benchmark was done and the ocean was taking more of the screen I mean, too hard to say exactly.

 

[tornadotsunamilife]

If I'm not mistaken AOE3 does HDR or another post render effect, pertty sure it was blur for the 6800's under sm 3.0, 6800's take a large hit from this so will the x800's if they do it too.

So you're saying using that specific path automatically enables a certain feature?

 

[razor1]

So you're saying using that specific path automatically enables a certain feature?

Exactly the paths really make a difference, Like for the sm 3.0 path for the far cry engine, parrallex bump mapping over rides normal mapping if the game or mod is being made for that in mind.

Pretty much when assigning shaders to materials on a mesh, the artists will assign the heighest level of shader which is designated for the heightest path, then the next lower path will do the same shader or lower shader depending on speed (this is all done automatically). Of course if the reviewer goes into the ini or cfg files and manually sets certain effects on then that will over ride this. Don't know if tom's did that or not.

Thats probably why they didn't put the x800's into the AOE benchmark, because of HDR and or blur. Again I played the demo when it first came out but can't remember which effect they used.

 

[tornadotsunamilife]

AFAIK isn't the HDR implemented in AoE3 FP16? That would explain the lack of it on X800's. However, isn't this what SCCT basically did with the sm3.0/sm1.1 (can't remember the exact path they used) , which left ati cards without parallax mapping (thank god they released a sm2.0 path in a patch). If so I find it most annoying

 

[Digital Viper-X-]

Exactly the paths really make a difference, Like for the sm 3.0 path for the far cry engine, parrallex bump mapping over rides normal mapping if the game or mod is being made for that in mind. Thats probably why they didn't put the x800's into the AOE benchmark, because of HDR and or blur. Again I played the demo when it first came out but can't remember which effect they used.

Well I have to say I'm playing AOE on my GTX @ 1600x1200 everything full except for no AA ^_^ runs pretty nice,

 

[Chris_Morley]

Show me a game where it was CRIPPLING it.

Well, it's conjecture since nobody is really producing the same effects under 3.0 as 2.0b, to my knowledge.

 

[razor1]

AFAIK isn't the HDR implemented in AoE3 FP16? That would explain the lack of it on X800's. However, isn't this what SCCT basically did with the sm3.0/sm1.1 (can't remember the exact path they used) , which left ati cards without parallax mapping (thank god they released a sm2.0 path in a patch). If so I find it most annoying

Yeah I know what ya mean, but Ubi is very close to nV, and there could have been some sctraching goin on

 

[John Reynolds]

Well, it's conjecture since nobody is really producing the same effects under 3.0 as 2.0b, to my knowledge.

Sure they are. Bethesda is one example of a dev saying that 3.0 is simply there for some performance gains but the game will look identical on 2.0 or 3.0 hardware.

 

[Chris_Morley]

Sure they are. Bethesda is one example of a dev saying that 3.0 is simply there for some performance gains but the game will look identical on 2.0 or 3.0 hardware.

That was the point of my post...what I'm saying is that developers are omitting effects if they give you the option of running in 2.0b or 3.0 for no other reason than performance. My post was an attempt to put it in perspective.

 

[Netrat33]

Well, it's conjecture since nobody is really producing the same effects under 3.0 as 2.0b, to my knowledge.

B&W2 does

 

[Netrat33]

Sure they are. Bethesda is one example of a dev saying that 3.0 is simply there for some performance gains but the game will look identical on 2.0 or 3.0 hardware.

Crytek said it too.

 

[Netrat33]

If I'm not mistaken AOE3 does HDR or another post render effect, pertty sure it was blur for the 6800's under sm 3.0, 6800's take a large hit from this so will the x800's if they did it too. Also they didn't test AOE 3 in that review for the x800's only on the 6800's did they test . . Yes keep rolling your eyes since that is all you can do.

So then you don't have any evidence or facts because HDR isn't an sm3 spec.

 

[John Reynolds]

That was the point of my post...what I'm saying is that developers are omitting effects if they give you the option of running in 2.0b or 3.0 for no other reason than performance. My post was an attempt to put it in perspective.

I think that's a bit of an assumption. Devs and their publishers have also been known to omit effects and/or features because of marketing deals they've made with IHVs.

 

[eno-on]

So then you don't have any evidence or facts because HDR isn't an sm3 spec.

I wasn't going to bring that up, because I wasn't sure. I don't remember HDR being part of sm3.0 spec either :\

 

[Netrat33]

I wasn't going to bring that up, because I wasn't sure. I don't remember HDR being part of sm3.0 spec either :\

But that's the thing, we still don't see any benchmarks between the two. Yes ATI cards couldn't do FP16 HDR before.

Splintercell chaos theory would be one example of benchmarking effects used in sm2.b and 3 (Turn off HDR!)

B&W2 we already see the hit really isn't a hit.