Tesselation done properly..according to AMD

[dust]

It's simple for me:

Does one desire tessellation in the future? Yes, no, maybe, don't care, the views may differ.

Does having more tessellation matter right now? Yes, no, maybe, don't care, the views may differ.

But, for me, it is always nice to see more performance based on there may be less of a tessellation bottleneck for titles. This adds some value to some but does it mean the end-all-be-all? Probably not.

But, when an IHV offers more performance than the other or may have some advantages, of course they're going to try to drive this point home. They wouldn't be doing their job. Just like the other is going to talk about balance, efficient, good enough, and attack why having too much is not the right way - they're doing their job.

But, since when is having too much performance of anything a bad thing?

Having more performance is never a bad thing, nobody here is saying that from what I can make out, but is it useful? There's the problem. I mean, it is marketed like it's the next best thing to fresh air but there aren't much titles to benefit from it yet, and most enthusiasts tend to buy their cards for games.

Everyone here is expecting it to pick up and become relevant, but, as of now it doesn't matter as much. The coming titles will probably have pretty demanding requirements that might render the current generation obsolete/weak, so all this "power" might prove insufficient. It's all about timing, NOT about downplaying progress.

Just ended a Metro session and believe me, I survived without DX11 and tessellation

 

[SirPauly]

It's all about singing the praises of strengths and downplaying weaknesses with PR to me. It's all about selling GPU's right now -- not yesterday or tomorrow. Perfect example is tessellation or x8 AA when nVidia was at a disadvantage.

End-users views are vast, subjective, tolerances, wallet size and passion -- that all differ - millions of views compared to two agendas that tries to create awareness and division to sell right now.

 

[extra]

Tesselation makes a difference now. And I will maintain this chant if the 6970 launches with superior tesselation power over a GTX580. Right about the time when everyone pulls a 180 who felt the opposite before.

Yep, the tessellation we have with the current cards makes a big difference and we are already at the point of diminishing returns imho! Cheers. We just need more software to use the darn thing!!!!!

 

[Keysplayr]

Yep, the tessellation we have with the current cards makes a big difference and we are already at the point of diminishing returns imho! Cheers. We just need more software to use the darn thing!!!!!

Might be nice if a few games utilized the Unigine Heaven 2.1 engine.
I wonder if anything like that is in the works.

 

[Modelworks]

The only thing tessellation does is allow developers to target a wider range of hardware performance levels. If everyone had the same hardware it wouldn't be a 'feature' . Tessellation can also cause more work for the developer. Some meshes do not tessellate well and require a lot of tweaking .

 

[Wreckage]

Wow, I really missed out on this one. Personally I think complaining about your competitor having too much performance, is not the best route. I think AMD should have just stayed silent on this issue. Just my 2 cents.

 

[NIGELG]

Wow, I really missed out on this one. Personally I think complaining about your competitor having too much performance, is not the best route. I think AMD should have just stayed silent on this issue. Just my 2 cents.

Yes they should have.But my 5850 is still doing a good job for me.The 570 looks good though.I'll wait to see what the 5950 does.

 

[SirPauly]

It's a tall order to convince someone that having less performance in a feature is a good thing. So, how does one try? What AMD is doing.

 

[NIGELG]

It's a tall order to convince someone that having less performance in a feature is a good thing. So, how does one try? What AMD is doing.

Show me a game which requires a lot of tessellation.Show me such a game which is unplayable on ATi's hardware.Show me a game which requires Nvidia's insane amounts of tessellation....now or in the year 2011.

 

[SlowSpyder]

Show me a game which requires a lot of tessellation.Show me such a game which is unplayable on ATi's hardware.Show me a game which requires Nvidia's insane amounts of tessellation....now or in the year 2011.

That's really what I've been saying in this thread. Nvidia has a better tessellator set up hands down. There is no arguing that less performance is better. The arguement is 'does it matter now or during the time that we are likely to use current hardware?' I typically keep my card about 18 months, I'm closing in on a year with this 5870. Many people here are enthusiasts who will jump to the next gen anyway.

With all that said, AMD is obviously going to have to improve performance in future parts as tessellation is used more.

 

[SirPauly]

2011 is not finished or even started yet. It's a simple point: Having more performance for a specific feature offers some value.

When AMD offered more performance with x8 AA, it was a good thing even though many claimed it wasn't needed, too much, over-kill, but for someone that appreciates things like this, it was nice to see because it offered value for some.

All I know is if tessellation is just fine or good enough then why on Earth is AMD and nVidia trying to improve upon it?

End-users don't buy their hardware at the same time or own them for the same time but when one has a performance advantage over a specific feature; it's a good thing because it may offer some value.

 

[Wreckage]

The only thing tessellation does is allow developers to target a wider range of hardware performance levels. If everyone had the same hardware it wouldn't be a 'feature' . Tessellation can also cause more work for the developer. Some meshes do not tessellate well and require a lot of tweaking .

This is part of the reason that you need not only a more powerful tessellation engine, but a more powerful GPU overall. Virtually every aspect of game programming can benefit from extra work by the developer, but we know they don't always do that. So you have to either hope for a patch or maybe the video card company will even have to step in and help (AA in the Unreal 3 engine).

I think AMD is painting itself into a corner with these statements. They could help developers to optimize their tessellation code, instead of complaining that NVIDIA is too powerful. Otherwise developers will just continue to target NVIDIA GPUs, just like with Physx, 3D, CUDA, AA in certain games, etc.

 

[Phynaz]

They could help developers to optimize their tessellation code

That's exactly what they did.

Their optimizations would have also benefited Nvidia.

 

[NIGELG]

2011 is not finished or even started yet. It's a simple point: Having more performance for a specific feature offers some value.

When AMD offered more performance with x8 AA, it was a good thing even though many claimed it wasn't needed, too much, over-kill, but for someone that appreciates things like this, it was nice to see because it offered value for some.

All I know is if tessellation is just fine or good enough then why on Earth is AMD and nVidia trying to improve upon it?

End-users don't buy their hardware at the same time or own them for the same time but when one has a performance advantage over a specific feature; it's a good thing because it may offer some value.

It offers zero value if games don't use it.I said 2011 because I wanted to know if there are any games coming in that year which will use this insane amount of tessellation.

 

[SlowSpyder]

2011 is not finished or even started yet. It's a simple point: Having more performance for a specific feature offers some value.

When AMD offered more performance with x8 AA, it was a good thing even though many claimed it wasn't needed, too much, over-kill, but for someone that appreciates things like this, it was nice to see because it offered value for some.

All I know is if tessellation is just fine or good enough then why on Earth is AMD and nVidia trying to improve upon it?

End-users don't buy their hardware at the same time or own them for the same time but when one has a performance advantage over a specific feature; it's a good thing because it may offer some value.

Right, it offers value if the feature is used during the cards lifetime. But you are assuming that tessellation is going to take off soon, or that tessellation will be a limiting factor. But that is just one aspect of the card. Did you realize that the Radeon 5870 has more than 60% more texture fillrate than the GTX480? What if that becomes a limiting factor? The problem is here that some are taking one feature and basing future games on that single feature. Other specs and features can come into play as well. Tessellation is stronger on the GTX480, we know that. But I doubt tessellation is going to be the single limiting factor in tomorrow games and we won't want to play them on current hardware as enthusiasts...

 

[SirPauly]

Imho,

Tessellation is just one aspect to consider and nothing more. Each individual has to figure out how important it is to them, just like other aspects. It's just nice to see an IHV raise the bar of tessellation and hopefully AMD and nVidia continue to do so.

 

[SirPauly]

It offers zero value if games don't use it.I said 2011 because I wanted to know if there are any games coming in that year which will use this insane amount of tessellation.

Games are using it now and a testament to AMD, nVidias and developers hard work. Extremer tests, right now, are more like synthetic benches to me that may gauge future forward abilities. Very important, but dismissed by some, imho.

 

[Rubycon]

nvm wrong thread!

 

[Aristotelian]

But I doubt tessellation is going to be the single limiting factor in tomorrow games and we won't want to play them on current hardware as enthusiasts...

And anyone with the tiniest bit of logical acumen has recognized that sort of reasoning and seen it in a number of posts for a while now. We use gaming benchmarks to understand how games play now. It's faulty reasoning to conclude that because x card is better at tessellation now (even though y card does it reasonably well too) that therefore x card will be better for most or all future games. End of story.

 

[SirPauly]

There isn't much logic when individuals use "we" in their posts.

 

[3DVagabond]

It offers zero value if games don't use it.I said 2011 because I wanted to know if there are any games coming in that year which will use this insane amount of tessellation.

We will never see games that use 1tri/pix. NURBS would be more efficient.

 

[Aristotelian]

There isn't much logic when individuals use "we" in their posts.

Quoting this for posterity regarding the lack of quality posting. There is no argument in your post, merely an insinuation. When you care to respond to the argument I presented, feel free to.

 

[SirPauly]

The points are simple: More tessellation performance may offer some value and may have less of a tessellation bottleneck in the future.

 

[ronnn]

The points are simple: More tessellation performance may offer some value and may have less of a tessellation bottleneck in the future.

So how much will a 580 be worth when the may comes true? I would predict very little, as all games I have seen are ports or as developers say multiplatform.

 

[Aristotelian]

The points are simple: More tessellation performance may offer some value and may have less of a tessellation bottleneck in the future.

Exactly, 'maybe'. Since it's 'maybe' it cannot necessarily count as a plus now. It's a 'maybe', not a 'will be', and that's all I have been trying to say. We're on the same page, then.