The ultimate question.

[Marvelous]

http://techreport.com/reviews/2004q3/radeon-x700xt/index.x?pg=3

In the wake of the Radeon X700's launch, some of you may have missed the news that ATI will be including application-specific optimizations in future revisions of its Catalyst drivers, including the ones we used for our X700 testing. I see there are a number of new articles kicking around out there today about this development, so I thought I'd point you to our coverage, as well. The first thing you need to do is read our article about a problem with Doom 3 texture filtering, in which we first reported that ATI would be incorporating app-specific optimizations. Next, go to this page of our Radeon X700 review to learn more about Catalyst A.I., including specific details about ATI's approach to optimizations and which games will be initially affected. We also have a quick write-up on Catalyst Control Center, ATI's new-look control panel for the Radeon series, which I think needs some additional work.
Once you understand what Catalyst A.I. is all about, hit this page of the review for DOOM 3 performance results on the X700 XT with and without Catalyst A.I. enabled. You'll see that ATI's optimizations make a big difference in performance. The scores on that page with the Catalyst 4.10 beta drivers are also directly comparable to the Cat 4.9 scores (without Catalyst A.I.) in our GeForce 6600 GT review. You'll notice that the Radeon X600 XT, 9800 XT, and X800 Pro also benefit greatly from the optimizations in the 4.10 beta drivers. For instance, the Radeon 9800 XT's score at 1024x768 jumps from 59.2 frames per second to 70.6. The primary reason for the performance increase, I believe, is a shader replacement that substitutes math operations in place of a texture value lookup for specular lighting, very similar to the famous hack at B3D. Doing the math is faster than accessing the texture on ATI's newer hardware, so the game runs faster. This is one way Cat A.I. can help ATI cope with the fact that's Doom 3's primary code path seems to be less than optimal for ATI hardware.

Catalyst A.I. also allows one to disable ATI's adaptive texture filtering, and we've demonstrated the mathematical differences between the adaptive and traditional approaches to texture filtering by subtracting one of our sample images from another for the benefit of you graphics eggheads out there.

There's the proof for CCC improving 10-20% at almost all resolutions.

 

[Marvelous]

Thats still a failure...you are not guaranteed shit when you flash or oc. you just contradicted your self...if you can't achive the XTPE speeds then what is the point of flashing? did you gain what you wanted? no. are the extra 4 pipes worth it? do you even KNOW the extra 4 pipes are even STABLE..have you ran them? there disabled for a reason. and it's usauly do to FAILED tests..but you still never showed me evidence of your claim for all the FACTS you have...

How am I contridicting myself when everyone who has gotten a x800 pro VIVO to flash it and get it to work. I mean WTF proof did you have.

And the VIVO's can be had for $400-420

 

[Marvelous]

Oh maybe you will be the first to justify to why i should buy a x800 that uses 2 year old tech, has no new tech, and has a hyped up core and ram...shit even intel makes new cores more than ati does.

LOL no new tech?

Longer shader programs — The X800's pixel shaders still have 24 bits of floating-point color precision for each color channel (red, green, blue, and alpha), and they do not have the branching and looping capabilities of the pixel shaders in the NV40. They do, however, have the ability to execute shader programs with longer instruction lengths, up from 160 in the 9800 XT to 1,536 in the X800. The X800's revised pixel shaders also have some register enhancements, including more temporary registers (up from 12 to 32). We'll look at the question of pixel shaders and shader models in more detail below.

3Dc normal map compression — Normal maps are simply grown-up bump maps. Like bump maps, they contain information about the elevation of a surface, but unlike bump maps, they use three-component coordinate system to describe a surface, with X, Y, and Z coordinates. Game developers are now commonly taking high-polygon models and generating from them two things: a low-poly mesh and a normal map. When mated back together inside a graphics card, these elements combine to look like a high-poly model, but they're much easier to handle and render.
Trouble is, like all textures, normal maps tend to chew up video memory, but normal maps don't tolerate well the compression artifacts caused by compression algorithms like DirectX Texture Compression (DXTC). If a normal map becomes blocky, the perceived elevation of a surface will become blocky and uneven, ruining the effect. ATI has tackled this problem by adapting the DTXC algorithm for alpha channel compression to work on normal maps. Specifically, the DXT5 alpha compression algorithm is used on the red and green channels, which store X and Y coordinate info, respectively. (Z values are discarded and computed later in the pixel shader.) This format is reasonably well suited for normal maps, and offers 4:1 compression ratios. Like any texture compression method, it should allow the use of higher resolution textures in a given amount of texture memory.



3Dc versus DTXC color compression. Source: ATI.

The X800 GPU supports this method of normal map compression, dubbed 3Dc, in hardware. Both DirectX and OpenGL support 3Dc via extensions, and game developers should be able to take advantage of it with minimal effort.

Temporal antialiasing — This new antialiasing feature is actually a driver trick that exploits the programmability of ATI's antialiasing hardware. We'll discuss it in more detail in the antialiasing section of the review.

http://techreport.com/reviews/2004q2/radeon-x800/index.x?pg=22

 

[bboynitrous]

Oh maybe you will be the first to justify to why i should buy a x800 that uses 2 year old tech, has no new tech, and has a hyped up core and ram...shit even intel makes new cores more than ati does.

Where is your justiffication of the smart model 3.0 in the 6800 yet? All those 2004 titles that were set to come out have not. 6800 is a very good card no doubt, but ATI was somewhat right about not needing SM3 right now. A year from now maybe it will, but until that time comes, both cards are damn fast and play everything. Even the x800. I don't need justification for my card, because I don't care enough. I'm happy because it's still fast and looks better than my geforce 3 and that is all that matters.

 

[fugu]

LOL no new tech? This shows how much of an Nvidiot you are.

keep up the name calling and you'll be banned

 

[CleanSlate]

Where is your justiffication of the smart model 3.0 in the 6800 yet? All those 2004 titles that were set to come out have not. 6800 is a very good card no doubt, but ATI was somewhat right about not needing SM3 right now. A year from now maybe it will, but until that time comes, both cards are damn fast and play everything. Even the x800. I don't need justification for my card, because I don't care enough. I'm happy because it's still fast and looks better than my geforce 3 and that is all that matters.

That guy you quoted doesn't know shit about the architecture or SM3... dont waste a post on an obvious lamer.

~Adam

 

[Marvelous]

There isn't any justification. XTPE beats an Ultra at highest resolutions with AA, ANiso.

Doom 3. Even then XTPE is very close to Ultra's am I right on this?

I repeat there's only a handful of OPENGL games. Almost all games are directX. Valve, Unreal, Farcry, All microsoft games These are all from different companies. Only one who makes opengl game is Activision and maybe EA when they borrow doom or quack engines.

XTPE has clear advantage on AA. You hook it up to your computer with the same monitor one by one and you'll notice this. Speed is where it's at and ATI has it.


And for you who hasn't gone to ATI child forums yet. http://hardforum.com/showthread.php?t=777427


Name calling my fault. I won't do it again. I would miss hardforum too much too.

 

[bboynitrous]

You are right Cleanslate, sorry, i'm just tired of people saying crap like that. It sickens me. Always crapping on other peoples cards. Wow, i'm way off topic. I try not to post in threads like this, but sometimes you can't resist

 

[bboynitrous]

6800GT is a great card. I would have bought one if they were out in time. X800 was out, fast and is nice overall. But now it's not as great of a buy. It's less risky to overclock than to flash/overclock.

 

[The Batman]

Ouch. I'm not even going to bother touching this one. Besides, I think the poll speaks loudly enough.

 

[Marvelous]

Most of people here voted because they already have a GT. I understand you want to justify what you bought. I for one don't have x800vivo or a GT. If it was up to me I would definitely go for a VIVO flashed to XT for sure. Every person in that thread mentioned above got there's to open up the pipes and work. Some of them couldn't overclock but still it's a sure thing. I'm not saying GT is not a great buy. It definitely is and it costs few bucks less and all the techno garbage Nvidia feeds little boys and girls with is sure enough for you to buy the card. But that's all PR if ask me.

I for one think that all x800pro Vivo cards are failed at x800XT speeds and they disabled the 4 pipes by bios. That is why the 4 yucky pipes are disabled. People who got their's open up the pipe could do it but some of them can not overclock at all. But when they have the pro bios in it overclocks to XT-PE levels with on 12 pipes of course.

Radeon 9500 to 9700 is different. These had 2 things wrong. First the pipes than the memory controller. There's 2 things wrong with this and that's basically half the card that is wrong with it because it's 4 pipes to 8 pipes. That is why there's fail rates.

X800pro VIVO is different. It has 256bit memory controller out of the box. It has only 1/4 of the pipes that probably failed at XT speeds with their stress test with stock cooling hance easy mod. If you have good cooling you won't have problems.

Non of the regular x800 pro have the extra 4 pipes.

I for one see it like this

x800 than GT than XT than Ultra, than XTPE

 

[Elios]

sigh i dont know whats up with some ppl i love my GT it run EVERYTHING i throw at and VERY fast at high res w/ AA and/or AF hell i can play CS:S at 60-100FPS @ 1280x1024 4xAA 8xAF and i get 75FPS in the stress test thats not slow to me >.> so what if the XT gets 5-10 more FPS im not going to see it and i get 12K in 3dmark03 and these are all DX
apps the GT is worth ever penny i spent and be for you going calling me and nVidia fan boy my last 2 cards were ATi a 9000 for a Athlon 900 and a 9600XT for this box both were great. IQ looks the same to me too i couldnt pick the 9600XT from the GT other then by FPS imo i just find that OCing a card "safer" then trying to flash to get 4 more pipes thay disable thoughs pipes for a reison and there not going to sell good 16pipe cores as 12pipe ones

 

[Terra]

Where is your justiffication of the smart model 3.0 in the 6800 yet? All those 2004 titles that were set to come out have not. 6800 is a very good card no doubt, but ATI was somewhat right about not needing SM3 right now. A year from now maybe it will, but until that time comes, both cards are damn fast and play everything. Even the x800. I don't need justification for my card, because I don't care enough. I'm happy because it's still fast and looks better than my geforce 3 and that is all that matters.

If there where no SM3.0 capable cards out there would come no SM3.0 games, simple as that...
First the hardware, then the games...

Terra...

 

[Elios]

oh Farcry has SM30 if you can find the 1.2 patch ^^ and thats DX as well ohmy

 

[Marvelous]

But is it faster than the XT with 4xAA 16x ANiso with those optimizations? Hells no so what is the point having your precious SM30? When Nvidia is not using SM30 nvidia's cards get beaten. If they have sm30 it's about even.

Although we could make out minor rendering differences between SM 2.0 and SM 3.0, speaking of reduced quality would be inappropriate. The intensity of the lighting may be slightly dimmer, but this is only noticeable in a direct side-by-side comparison of screenshots.

As was to be expected, SM 3.0 does not improve rendering quality, even though NVIDIA has repeatedly suggested just that in several of its presentations. Whether the results we saw here in FarCry will translate into a general advantage in SM 3.0 games for NVIDIA remains to be seen, as the performance improvements could be caused by other factors as well. For example, it is entirely possible that the new SM3.0 code path also contains shader code specifically optimized for the NV4x architecture.

Despite this improvement in NVIDIA's newest platform, we remain critical of the performance and image quality that the GeForce FX 5950 Ultra offers. Although the image quality showed an overall improvement, it is still not on par with that of the newer GeForce models or ATi's Radeons. On top of that, some of the results we saw in this comparison are downright disappointing compared to a Radeon 9800 XT. Perhaps this was caused by the v61.45 driver, which is not specifically optimized for the GeForce FX 5950 Ultra. On the other hand, maybe this is just the price you have to pay for improved image quality. It's hard to say.

SM 3.0 does not improve rendering quality, even though NVIDIA has repeatedly suggested just that in several of its presentations. LOL

http://graphics.tomshardware.com/graphic/20040705/farcry-12.html

 

[Terra]

But is it faster than the XT with 4xAA 16x ANiso with those optimizations? Hells no so what is the point having your precious SM30?

FraCry has some SM3.0 features, but is not build as a SM3.0 game.
Wait until native SM3.0 games comes out...
And I will save your posts for when when the R520(wich should be SM3.0) comes out, and then repost them...
Then it will be fun to laugh at your posts and see if you still got this attitude then, or you then will praise SM3.0

Terra...

 

[Marvelous]

than why do people even mention Farcry?

By the time those SM30 games come out these cards will be useless like the Geforce 2 or 3 for that matter. When was these cards released? Like couple months ago. It takes years to make a good game.

You can tell the future? You are not a fortune teller. If you were you wouldn't be here trying to tell the future. You would invest lots of money make billions. Who knows what will happen. What makes you think SM20 won't be supported in these future games. It might make a slight difference but then it can't cope with fillrate, bandwidth it just can't push frames. Only time will tell.

 

[Elios]

But is it faster than the XT with 4xAA 16x ANiso with those optimizations? Hells no so what is the point having your precious SM30? When Nvidia is not using SM30 nvidia's cards get beaten. If they have sm30 it's about even.



SM 3.0 does not improve rendering quality, even though NVIDIA has repeatedly suggested just that in several of its presentations. LOL

http://graphics.tomshardware.com/graphic/20040705/farcry-12.html

16XAF is just useless i have a hard time telling 4x from 8x as it is >.> but ill run a bench mark at 16x bet its not much slower >.> not that 16xAF is much better then 8 imo

 

[Marvelous]

16XAF is just useless i have a hard time telling 4x from 8x as it is >.> but ill run a bench mark at 16x bet its not much slower >.> not that 16xAF is much better then 8 imo

Well that's you. Just because you can't tell the difference don't mean other people can't. I mean why buy these $400 cards if you can't crank up the resolution and use every feature? Faster CPU's will be out and about and more faster GPU's. You buy what is current not for the future. Doom 3, Farcry, Half Life, Splinter Cell, Unreal. These are the game engines that will be run on these cards. Not some Nvidia's SM30 tech demos.

IQ looks the same to me too i couldnt pick the 9600XT from the GT other then by FPS imo i just find that OCing a card "safer" then trying to flash to get 4 more pipes thay disable thoughs pipes for a reison and there not going to sell good 16pipe cores as 12pipe ones

How you been reading. ALL VIVO X800pro owners got there Pipes to work. What more proof do you need? http://hardforum.com/showthread.php?t=777427

 

[rancor]

http://techreport.com/reviews/2004q3/radeon-x700xt/index.x?pg=3





There's the proof for CCC improving 10-20% at almost all resolutions.

Thats proof for only Doom 3, if you look at UT 2004, CCC actually slowed it down.

It has varing effects as of now depending on games. It might end up over all speeding things up, but not by much. Since shaders are compiled at run time anyways, they kinda are already card specific, since the compiling is taken care of by the specific card being used. You really can't get much more speed out of a shader unless its completely changed, AKA doom, replacing a texture lookup with a mathmatical approximation. And when this occurs its very noticable. As 4.9 drivers showed there was a slight IQ drop, Not saying it looked bad, but it looked a bit different, as specular highlights were a bit more etc.

 

[creedAMD]

Most of people here voted because they already have a GT. I understand you want to justify what you bought.

Why do think so many people bought a GT? It's not like we don't know what reviews are. Hell we're on a hardware review site, ffs. I suppose that you don't remember what it was like around here when the r300 was laying the smack down? I guess everyone was just justifing their purchase then, too? I am a big fan of my 9800pro, 9700pro, 9500pro, but I bought the 6800gt because I still like playing opengl games, I don't care how many there are, if I want to play one opengl game, I know that I will be getting a hell of lot more noticable performance gain than if I was playing on an Ati card, the same can not be said about ANY directx games. NONE. You would not be able to tell the difference on any dx games.

 

[rancor]

than why do people even mention Farcry?

By the time those SM30 games come out these cards will be useless like the Geforce 2 or 3 for that matter. When was these cards released? Like couple months ago. It takes years to make a good game.

You can tell the future? You are not a fortune teller. If you were you wouldn't be here trying to tell the future. You would invest lots of money make billions. Who knows what will happen. What makes you think SM20 won't be supported in these future games. It might make a slight difference but then it can't cope with fillrate, bandwidth it just can't push frames. Only time will tell.

How could you say they will be useless, most games that just came out are still runnable on a 9700, most games are backward compatible for 2 years at least, damn Doom 3 still works just fine on a gf 2 mx. UT 2004 runs just fine on a geforce 3, just at lower resolutions. Unreal 3 is going to be fully playable on this generation of cards, not with everything turned up but it will be playable. Far Cry works fine on a geforce 4 ti.

 

[rancor]

Is there something you want to add?

This thread was about choosing which vid card x800 flash to xtpe or GT to Ultra Speeds. I stated my opinion from the articles I have read here, techreport, anandtech. XTPE does wipe a GT or even an Ultra for that matter. Look at the real resolutions you will be playing 1600x1200 4x AA, 16x aniso I presume if your going to shell out $400+ on a vid card.

All VIVO x800 is a guaranteed. There's not 1 failure. I've read every post for post. Some of them couldn't get to XTPE speeds but they still had all 16 pipes to work.

And my child hood? Okay whatever you say. You don't know nothing about me.

They don't wipe the board clean, thats the point, all of them still use that 61.77 drivers, with no aa and af on the the nV cards are in the lead, and nV did optimize thier aa and af since that driver release. And they are still working on it. Thats why they are equal. I'm not going to say nV is faster, intruth they might end up there, but we don't know how far they will take thier aa and af opts. ATi does some of thier aa and af opts within hardware itself like angle dependent af. And these opts can't be turned off by CCC

 

[Token User]

Why was there no option for the S3 Trio Virge card? Its PCI and everything! (j/k) ....

Ultimately, it sounds like it comes down to $$$, otherwise instead of buying lower and flashing/OC'ing to the higher model, you'd just be buying the higher end card.

Build your ultimate rig now - but if you are interested in HL/2 ... its not out yet. No benches on whcih card will perform better. Wait ... look at the benches ... then decide. New technology is always coming out. Live with the existing card you have until YOU are comfortable with the decision, then do it - don't come looking for validation ...it will degenerate into a flame fest. Post your benches - let the results speak for themselves.

 

[CastleBravo]

Lots of people go from 12 to 16 pipes flashing a VIVO x800 Pro just fine. Lots of people also FAIL to get x800 XT-PE clock speeds out of them. Why? Because ATI aren’t retards. They aren’t going to build an x800 XT-PE, then sell it for less money than an x800 XT-PE just because they want to make you feel like a 1337 d00d when you turn on the 4 disabled pipelines. Some people get a free ride because they get lucky and their card can take the higher clock speeds, some people don’t. There are lots of topics on other forums bitching about how their x800 Pro VIVO->x800 XT-PE flashes “worked,” but they still can’t get near XT-PE clocks.

The end result of this is that a lot of these people are basically getting an underclocked x800 XT-PE for their trouble, which really only puts them at parity with the 6800 GT and cost them the same amount of money in the first place. And of course voided their warranty in a rather obvious way, if you care about that sort of thing.

The 6800 GT is NOT a sure thing to run at Ultra speeds, either. Some of these things just run too hot in some people’s PCs. But even IF it doesn’t OC for crap, it still is a 16-pipeline card with SM 3.0 and performance shockingly close to the 6800 Ultra in most applications. You still have a card that beats even the x800 XT-PE at Doom 3, and rather handily at that. You still have a card that beats its counterpart, the x800 Pro, at almost everything, and essentially ties everything else. Heck, you can run Far Cry with the 1.2 patch, DX9.0c and enable the SM 3.0 path to beat the crap out of ATI at its own game… so much for ATI superiority with DX9 games.

 

[darktiger]

I vote 6800GT --> Ultra so you can enjoy the benefits now....

 

[[RIP]Zeus]

sorry i didn't get back to you last night marv...lol
i went to bed cause i have work. anyways it seems everyone here just showed the proof you needed from me to prove you wrong. but if you need more plz tell me..also why haven't you showed me proof of your claim's?
i am still waiting. plz do not disapoint me.....

 

[Marvelous]

How could you say they will be useless, most games that just came out are still runnable on a 9700, most games are backward compatible for 2 years at least, damn Doom 3 still works just fine on a gf 2 mx. UT 2004 runs just fine on a geforce 3, just at lower resolutions. Unreal 3 is going to be fully playable on this generation of cards, not with everything turned up but it will be playable. Far Cry works fine on a geforce 4 ti.

Do you seriously think x800 cards won't be supported as much as say an 6800 card is outright rediculous. ATI has a way of rendering graphics and so does GT. You my friend just look at specs and say that's bettter that one. Look at the big picture. Game makers will make it suitable to run on the lowest card which will be in the future are Radeon 9500-9700 series FX series because these were the first cards to have pixel and shaders. These future games your referring to are based on what? Doom 3 engines, Unreal engines, Half Life, Cryotek, Splinter Cell, etc etc...

An XT still outdoes a GT anyday.

I've told you many times I like the GT if VIVO wasn't available I would get the GT myself.

They don't wipe the board clean, thats the point, all of them still use that 61.77 drivers, with no aa and af on the the nV cards are in the lead, and nV did optimize thier aa and af since that driver release. And they are still working on it. Thats why they are equal. I'm not going to say nV is faster, intruth they might end up there, but we don't know how far they will take thier aa and af opts. ATi does some of thier aa and af opts within hardware itself like angle dependent af. And these opts can't be turned off by CCC

IF your talking about Temporal AA for Radeon cards. Yes this can be turned off. Also ATI has the option to use Bilinear and AF for more performance without no noticeable image degrading. And ATI way of doing Aniso angle dependent AF is wrong? Okay whatever too much techno garbage eating out your ears.

[RIP]Zeus sorry i didn't get back to you last night marv...lol
i went to bed cause i have work. anyways it seems everyone here just showed the proof you needed from me to prove you wrong. but if you need more plz tell me..also why haven't you showed me proof of your claim's?
i am still waiting. plz do not disapoint me.....

You know I won't dissapoint. What do you mean I didn't show you any proof? Links were provided to benchmarks where XT hurts a GT. x800pro VIVO flash to XT

Lots of people go from 12 to 16 pipes flashing a VIVO x800 Pro just fine. Lots of people also FAIL to get x800 XT-PE clock speeds out of them. Why? Because ATI aren’t retards. They aren’t going to build an x800 XT-PE, then sell it for less money than an x800 XT-PE just because they want to make you feel like a 1337 d00d when you turn on the 4 disabled pipelines. Some people get a free ride because they get lucky and their card can take the higher clock speeds, some people don’t. There are lots of topics on other forums bitching about how their x800 Pro VIVO->x800 XT-PE flashes “worked,” but they still can’t get near XT-PE clocks.

The end result of this is that a lot of these people are basically getting an underclocked x800 XT-PE for their trouble, which really only puts them at parity with the 6800 GT and cost them the same amount of money in the first place. And of course voided their warranty in a rather obvious way, if you care about that sort of thing.

The 6800 GT is NOT a sure thing to run at Ultra speeds, either. Some of these things just run too hot in some people’s PCs. But even IF it doesn’t OC for crap, it still is a 16-pipeline card with SM 3.0 and performance shockingly close to the 6800 Ultra in most applications. You still have a card that beats even the x800 XT-PE at Doom 3, and rather handily at that. You still have a card that beats its counterpart, the x800 Pro, at almost everything, and essentially ties everything else. Heck, you can run Far Cry with the 1.2 patch, DX9.0c and enable the SM 3.0 path to beat the crap out of ATI at its own game… so much for ATI superiority with DX9 games.

The point of this thread is for new people who hasn't gotten a GT or a VIVO. You said it yourself all VIVO cards softmods to XT speeds easy and most of them go beyond that.

GT against a VIVO XT is no brainer. XT wipes GT. That's for sure with or without SM30. You know why right? XT and XTPE is 1 gigantic fillrate monsters.

 

[rancor]

Do you seriously think x800 cards won't be supported as much as say an 6800 card is outright rediculous. ATI has a way of rendering graphics and so does GT. You my friend just look at specs and say that's bettter that one. Look at the big picture. Game makers will make it suitable to run on the lowest card which will be in the future are Radeon 9500-9700 series FX series because these were the first cards to have pixel and shaders. These future games your referring to are based on what? Doom 3 engines, Unreal engines, Half Life, Cryotek, Splinter Cell, etc etc...

An XT still outdoes a GT anyday.

I've told you many times I like the GT if VIVO wasn't available I would get the GT myself.

You know I won't dissapoint. What do you mean I didn't show you any proof? Links were provided to benchmarks where XT hurts a GT. x800pro VIVO flash to XT


If your talking about Temporal AA for Radeon cards. Yes this can be turned off. Also ATI has the option to use Bilinear and AF for more performance without no noticeable image degrading. And ATI way of doing Aniso angle dependent AF is wrong? Okay whatever too much techno garbage eating out your ears.


The point of this thread is for new people who hasn't gotten a GT or a VIVO. You said it yourself all VIVO cards softmods to XT speeds easy and most of them go beyond that.

GT against a VIVO XT is no brainer. XT wipes GT. That's for sure with or without SM30. You know why right? XT and XTPE is 1 gigantic fillrate monsters.

I'm sorry the gf 2's were the first to start using vertex shaders, gf 3 where the first to use pixel shaders and vertex shaders followed by the 8500's, not the fx and 9700 series these lines were the first to use sm 2.0.

No I wasn't talking about temporal AA, thats not an optimization, see there ya go, you talk about things you don't even know the definition of let alone understand.

Btw you are saying your vivo is overclocked with all 16 pipelines, why are you trying to compare that with a regular GT, not an overclocked to ultra speeds?

Vivo's do softmod, but to get xt pe speeds isn't as easy as gt to ultra speeds.

Also as much as games will be backward compatiable it would be nice to use new features when they are available, ps 2.0b won't be used as much as sm 3.0. Thats pretty obvious, because ATi is coming out with an sm 3.0 part themselves. Why would developers make a seperate path, thats kinda hard work, with amount of shaders that have to be rewritten to take advantage of that path.

 

[CastleBravo]

You said it yourself all VIVO cards softmods to XT speeds easy and most of them go beyond that.:

Um, no I didn't. Try reading, it can be helpful.

I said they flash from Pro VIVO to XT fine, but often ***DON'T*** hit XT-PE clock speeds. There are lots of topics talking about this on other forums, the flash isn't the problem, the problem is getting the clocks of the XT-PE to run stable on a Pro Vivo.

"XT still outdoes a GT anyday"

But not your Pro VIVO if it can't hit GT speeds. Or the GT clocks faster than a 6800 Ultra.

 

[Marvelous]

I'm sorry the gf 2's were the first to start using vertex shaders, gf 3 where the first to use pixel shaders and vertex shaders followed by the 8500's, not the fx and 9700 series these lines were the first to use sm 2.0.

No I wasn't talking about temporal AA, thats not an optimization, see there ya go, you talk about things you don't even know the definition of let alone understand.

Then where is your geforce 2 vertex shader performance or even geforce 3 pixel shader performance for that matter? In your logice Geforce 2 or 3 will run these newer games but it really doesn't. If you call 20fps in low quality acceptable that's you.

XT does it with pure fillrate power and memory bandwidth GT does SM30 with lower fillrate. Does this mean way ATI way of rendering 3D or Nvidia way wrong? No it does not mean that. ATI rather push more fillrate and bandwidth to even outstand sm30. That's the whole point I'm trying to make here. 6800GT gets beaten up when SM30 is not used. Does SM30 give you better picture quality? No it doesn't as the link I provided few posts up.

Yeah whatever I don't know what I'm talking about.

Why don't you make it clear for people to understand instead of mumbling. FYI temporal AA is an optimization. It clears pictures and minimal or no loss in performance.

Btw you are saying your vivo is overclocked with all 16 pipelines, why are you trying to compare that with a regular GT, not an overclocked to ultra speeds?

Vivo's do softmod, but to get xt pe speeds isn't as easy as gt to ultra speeds.

I'm saying this again for the 3rd time. I do not have a radeon x800pro vivo. I wish I did.

Why am I trying to compare to a GT? Because this whole post was based on this. Look at the pole above.

6800GT -> Ultra Speed
x800Pro -> Flash / XT Speed
Wait, get x800XT later

But most of them are getting near or close to XT-PE speeds. I'm sure not everyone who has GT is not getting ultra speeds. I gave you the link to prove that all vivo's do softmod. maybe you bought your precious card on sale for good price. Currently I have radeon 9700pro and geforce ti4200 on my backup machine. I still haven't made the purchase becaue I do not like to stay in bleeding edge with graphics card and these prices are freaking rediculous just for a card.

Also as much as games will be backward compatiable it would be nice to use new features when they are available, ps 2.0b won't be used as much as sm 3.0. Thats pretty obvious, because ATi is coming out with an sm 3.0 part themselves. Why would developers make a seperate path, thats kinda hard work, with amount of shaders that have to be rewritten to take advantage of that path.

You don't think they would rather use the lowest common denomintor. I would probably start with Radeon 9700pro if I was a coder. Wouldn't you? Or do you think they'll just exculusively use SM30. LOL If you don't have fillrate and bandwidth these sm30 or whatever is useless. By the time these sm30 games start to appear we will be on PCI Express running sli enjoying life. SM30 is nothing more than improving performance if used like FarCry for instance. While it is a patch Cryotek themselves worte the patch not some coder in his basement. This will most likely the way SM30 will impact on current games and future games to come. As say Doom 3 as well. It takes years to make a game engine. Quake 3 engine was used for so many years so did Unreal engine until it evolved to unreal 2 engine.

Again I explained why ATI wins. Fillrate, Bandwidth. A GT is clocked at what 350mhz. An ultra clocked at 400mhz. A XT is clocked what 500mhz. Do the math.

 

[Mastericy]

The ultimate question is what game you want to play. 6800 is the card for Doom 3, X800 for HL2. ( If it ever gets here )

 

[rancor]

Then where is your geforce 2 vertex shader performance or even geforce 3 pixel shader performance for that matter? In your logice Geforce 2 or 3 will run these newer games but it really doesn't. If you call 20fps in low quality acceptable that's you.

XT does it with pure fillrate power and memory bandwidth GT does SM30 with lower fillrate. Does this mean way ATI way of rendering 3D or Nvidia way wrong? No it does not mean that. ATI rather push more fillrate and bandwidth to even outstand sm30. That's the whole point I'm trying to make here. 6800GT gets beaten up when SM30 is not used. Does SM30 give you better picture quality? No it doesn't as the link I provided few posts up.

Yeah whatever I don't know what I'm talking about.

Why don't you make it clear for people to understand instead of mumbling. FYI temporal AA is an optimization. It clears pictures and minimal or no loss in performance.


I'm saying this again for the 3rd time. I do not have a radeon x800pro vivo. I wish I did.

Why am I trying to compare to a GT? Because this whole post was based on this. Look at the pole above.

6800GT -> Ultra Speed
x800Pro -> Flash / XT Speed
Wait, get x800XT later

But most of them are getting near or close to XT-PE speeds. I'm sure not everyone who has GT is not getting ultra speeds. I gave you the link to prove that all vivo's do softmod. maybe you bought your precious card on sale for good price. Currently I have radeon 9700pro and geforce ti4200 on my backup machine. I still haven't made the purchase becaue I do not like to stay in bleeding edge with graphics card and these prices are freaking rediculous just for a card.





You don't think they would rather use the lowest common denomintor. I would probably start with Radeon 9700pro if I was a coder. Wouldn't you? Or do you think they'll just exculusively use SM30. LOL If you don't have fillrate and bandwidth these sm30 or whatever is useless. By the time these sm30 games start to appear we will be on PCI Express running sli enjoying life. SM30 is nothing more than improving performance if used like FarCry for instance. While it is a patch Cryotek themselves worte the patch not some coder in his basement. This will most likely the way SM30 will impact on current games and future games to come. As say Doom 3 as well. It takes years to make a game engine. Quake 3 engine was used for so many years so did Unreal engine until it evolved to unreal 2 engine.

Again I explained why ATI wins. Fillrate, Bandwidth. A GT is clocked at what 350mhz. An ultra clocked at 400mhz. A XT is clocked what 500mhz. Do the math.

Your posts are like swiss cheese so many holes, and you still don't have a clue when its pointed out.

God could be any more blind, I'm not even going to get into semantics with you, you talk so much fud, sound like ATi pr

Tell me why sm 3.0 will hurt fillrate and bandwidth

Temporal aa is not an optimization, its actaully the same speed as regular AA, but you you need a minimal of 60 fps. Lets see if I can explain this in simple terms. Temporal AA, inceases AA perfomance by switching between the different AA modes as necessary, but if you don't have at least 60 fps (the human eye can't really notice AA while playing games at this frame rate) there is a blurring effect, because of the AA levels switching.

Do you know shaders were there in the first cards avaiable? yeah thought so, Just that the gf 2's started with programmable (by programmable I mean easy to access, in truth you could still program to shaders before hand but you would have had to go though some custom drivers, and hand written ASM. Gf 3 is fully programmable, and there is alot you can do with them, just as if you tried to play doom 3 on it it works just fine just have go to lower resolutions. As with Far Cry, I played Far Cry on a 866 with a gf 3 ti 500 with no problem at all, low res but it worked just fine

If you don't have the card how can you recommend it? ok Marv what ever.

Most of them are getting XT PE speeds are you sure? show me the links.

No I always start of with the highest setting first, then code to lower end cards, for stibility issues, Since the new features require more flexiblity in the core, SINCE YOU ARE NOT A PROGRAMMER you would not understand this. This is the process, the more complex a shader or program, that can be created the more chances of it not being able to do that shader or program on a lower end system, so you start off with what you know is possible then work your way back and do specific optimizations so you can get it working on older tech. If the tech isn't available to the programmer yes then ya go through the math of what you are trying to do and side step the parts that will have issues with lower end cards, but this is not easy, its much harder because there is no base, you are going by the math and not the code. If you try to optimize before a certain point the code gets too messy for even the orginal programmer to understand. This is because optimized code doesn't follow any specific guidelines unlike the orginal code.

Why don't you show us the math on fillrate, shader performance, and bandwidth, what parts are neglegible depending on different situations? I can go through this. since you brought it up enlighten us.

Things like SM 3.0 I don't need to recode awhole lot to add it in, since switching from dx9 to dx9c is a 2 or 3 lines and a recompile. Even easier in Ogl, just a recompile. (of course have to add in shaders that support sm 3.0 though)

Unreal 2 to 3 is a big step because the previous engine didn't have the proper support for shaders (engines have to be made a certain way for different shaders). Look at Doom 3, there is already a mod (yes its a mod not made by Id) that has Parrellax bump mapping. its not hard do as long as the engine supports shaders. The Cry engine now supports sm 3.0 at least the version we are using. It didnt' take long for them to add it in.

 

[Marvelous]

Your posts are like swiss cheese so many holes, and you still don't have a clue when its pointed out.

God could be any more blind, I'm not even going to get into semantics with you, you talk so much fud, sound like ATi pr

Tell me why sm 3.0 will hurt fillrate and bandwidth

Temporal aa is not an optimization, its actaully the same speed as regular AA, but you you need a minimal of 60 fps. Lets see if I can explain this in simple terms. Temporal AA, inceases AA perfomance by switching between the different AA modes as necessary, but if you don't have at least 60 fps (the human eye can't really notice AA while playing games at this frame rate) there is a blurring effect, because of the AA levels switching.

Do you know shaders were there in the first cards avaiable? yeah thought so, Just that the gf 2's started with programmable (by programmable I mean easy to access, in truth you could still program to shaders before hand but you would have had to go though some custom drivers, and hand written ASM. Gf 3 is fully programmable, and there is alot you can do with them, just as if you tried to play doom 3 on it it works just fine just have go to lower resolutions. As with Far Cry, I played Far Cry on a 866 with a gf 3 ti 500 with no problem at all, low res but it worked just fine

If you don't have the card how can you recommend it? ok Marv what ever.

Most of them are getting XT PE speeds are you sure? show me the links.

No I always start of with the highest setting first, then code to lower end cards, for stibility issues, Since the new features require more flexiblity in the core, SINCE YOU ARE NOT A PROGRAMMER you would not understand this. This is the process, the more complex a shader or program, that can be created the more chances of it not being able to do that shader or program on a lower end system, so you start off with what you know is possible then work your way back and do specific optimizations so you can get it working on older tech. If the tech isn't available to the programmer yes then ya go through the math of what you are trying to do and side step the parts that will have issues with lower end cards, but this is not easy, its much harder because there is no base, you are going by the math and not the code. If you try to optimize before a certain point the code gets too messy for even the orginal programmer to understand. This is because optimized code doesn't follow any specific guidelines unlike the orginal code.

Why don't you show us the math on fillrate, shader performance, and bandwidth, what parts are neglegible depending on different situations? I can go through this. since you brought it up enlighten us.

Things like SM 3.0 I don't need to recode awhole lot to add it in, since switching from dx9 to dx9c is a 2 or 3 lines and a recompile. Even easier in Ogl, just a recompile. (of course have to add in shaders that support sm 3.0 though)

Unreal 2 to 3 is a big step because the previous engine didn't have the proper support for shaders (engines have to be made a certain way for different shaders). Look at Doom 3, there is already a mod (yes its a mod not made by Id) that has Parrellax bump mapping. its not hard do as long as the engine supports shaders. The Cry engine now supports sm 3.0 at least the version we are using. It didnt' take long for them to add it in.

Tell me why sm 3.0 will hurt fillrate and bandwidth

I never said this? I said it will improve performance and it does. I also said XT cards have more fill rate to outstand SM30 than say a GT card.

Temporal AA does not increase performance. Do you even read articles? It basically makes lines more smoother by putting more samples into edges.

Reality is XT beats a GT hands down. The poll was about buying which card. It's obviously a no brainer they are very close in price. I for one rather have the sheer brute strength of x800pro----------->XT+ vivo features. I mean just look at the fill rate difference? That's almost 50% more fillrate. 350mhz vs 500mhz. Small memory bandwidth or big memory bandwidth?

Okay how long did Unreal engine come out? 1998 When did unreal 2 engine game come out? 2002 Four years. By the time it takes carmack to make another engine will be 3-4 years by that time these cards will be useless. Carmack just made Doom and finished unless somebody pulls a magic trick there's nobody who codes for OPENGL. Who else makes OpenGL games can you name some? I'm not talking about companies who borrow quake 3 or doom engines.

There's no point arguing with you it's just going to be like my GT is better than x800pro vivo flashed to XT.

I don't know what going to choose I'm just going to wait it out. I can already play any game I throw. So I'm not in a rush but those VIVO does look mighty scrumtious.

 

[rancor]

I never said this? I said it will improve performance and it does. I also said XT cards have more fill rate to outstand SM30 than say a GT card.

Temporal AA does not increase performance. Do you even read articles? It basically makes lines more smoother by putting more samples into edges.

Reality is XT beats a GT hands down. The poll was about buying which card. It's obviously a no brainer they are very close in price. I for one rather have the sheer brute strength of x800pro----------->XT+ vivo features. I mean just look at the fill rate difference? That's almost 50% more fillrate. 350mhz vs 500mhz. Small memory bandwidth or big memory bandwidth?

Okay how long did Unreal engine come out? 1998 When did unreal 2 engine game come out? 2002 Four years. By the time it takes carmack to make another engine will be 3-4 years by that time these cards will be useless. Carmack just made Doom and finished unless somebody pulls a magic trick there's nobody who codes for OPENGL. Who else makes OpenGL games can you name some? I'm not talking about companies who borrow quake 3 or doom engines.

There's no point arguing with you it's just going to be like my GT is better than x800pro vivo flashed to XT.

I don't know what going to choose I'm just going to wait it out. I can already play any game I throw. So I'm not in a rush but those VIVO does look mighty scrumtious.

Right and 3d engine development has been much faster since Quake 1, 2, 3, it took Crytek 2 years to develop thier engine, there is another engine coming out that is just as capable as Unreal 3, called the Reality engine (actaully its already being used to make a game the engine is done), Then you have Stalker. Unreal 2003, then Unreal 2004 same engine one was dx 8 and one was dx 9, one year apart, Unreal 3 next year 2 years from Unreal 2004, Carmack can't just sit on his ass anymore, In the first 5 years of RT 3d engines there hasn't been much difference in technology, in the last 3 years there has been plenty more. But each new Dx and Ogl, its more of changing of few things here and there, its easier to add in the changes.

Why are you talking about the GT and not a GT at ultra speeds?

The poll was about a gt to ultra overclock vs, pro vivo to xt mod go back and look at the thread title and poll!

Temporal AA does not increase performance. Do you even read articles? It basically makes lines more smoother by putting more samples into edges.

Yes I gave you a indepth explination of how it does it and I never read that article, because I don't need to

And where are your numbers? Why won't you produce them Marv?

 

[Marvelous]

What are numbers are you keep talking about? In Farcry Geforce 6800ultra with the 1.2 patch still gets beaten by the XT. You want a link? http://www.anandtech.com/video/showdoc.aspx?i=2102&p=5


And Most ViVO flashed cards go beyond XT speeds. XT is the minimal speed you get out of the box.

 

[rancor]

What are numbers are you keep talking about? In Farcry Geforce 6800ultra with the 1.2 patch still gets beaten by the XT. You want a link? http://www.anandtech.com/video/showdoc.aspx?i=2102&p=5


And Most ViVO flashed cards go beyond XT speeds. XT is the minimal speed you get out of the box.

yeah ok 61.45 nice try there.

And I want the fillrate, shader performance, numbers. That matter more because there are no new benchmarks with the newer drivers yet, btw we opened a can of worms

 

[Marvelous]

Why don't you post it? Since your such a smarty pantz. I mean where's this review? Where does a GT beat a XT? Show me where.

 

[rancor]

Why don't you post it? Since your such a smarty pantz. I mean where's this review? Where does a GT beat a XT? Show me where.

Since you raised it I want you to show it to me
I'm not saying the gt beats the xt. They are close, the ultra beats the xt, but matchs the xt pe. Thats what I've been saying. lol

 

[Marvelous]

WTF I showed you where XT beats a GT. Show me where the 6800GT beats a XT? You have no evidence no proof nothing. You just keep saying your 61.47 or whatever. I don't care it's probably a game specific optimization like quack. That is why ATI will be doing the same thing hence CCC.

But yeah what i've been saying is that this Vivo goes beyon XT speeds easy more likely XTPE speeds so ATI wins. BOO HOO LOL