[AMD Tessalation] Is This True score?

[Massman]

Well, it's just a DX11 benchmark. Tesselation is just one of the things.

[wesjuhdabomb]

So... If i make a score now with my 5770 and i put Tesselation on OFF (0x) it will still be legit now? Since we are talking about bugged scores now and a while ago in this topic it was mentioned that is was no problem to put tesselation to OFF.

Soo... If there will come some check for this, will my scores posted now still be legit?
Botomline, is putting tesselation to OFF legit now or not?

[|ron]

Someone made a comparative screenshot of unigine with tessellation disabled from the benchmark and with tessellation at 0x in the CCC: the result was very different... even if we use 0x in the CCC, there's still some calculations to be made for a minimum level of tessellation.
Now, if all this was bs, then we must take measures... if everything it's correct then I see it as a normal tweak, just like setting mipmap quality on performance, changing the LODs etc...

[Vinster]

This is wacked... but as one already put it, it's like the LOD adjustment and I see it the same way. I've played with LOD in 3DM01 and I got a better score but really wasn't worth watching. I see it as the GPU not having to work as hard and being able to yield a better score.

To keep it fair, note it as a tweak in the bench rules so everyone can use it. and I would recommend the same condition for 3D01 and 03 with LOD on nVidia cards.

If everyone does this available tweak then there is no off-level playing field. and AMD gets their well deserved turn at owning on a bench as nVidia has and does for 3DM01 and 03.

Just my 2 cents.

Vin

PS: I don't benefit from this, I only now own nVidia cards. My AMD cards have been gone for a bit, nor do I plan on getting any in the near future.

[Computurd]

the hidden river in crysis 2 is epic. its under the street cant even see it. LOL invisible tesselation everywhere. set to favor nvidias cards.
AMD fires back! Turn that crap off L:OL

[RagingCain]

Originally Posted by chew*

Yes that is the reason.

For reference AMD reccomends the slider in the middle which is 32X iirc for a balance of good performance with adequate tesselation but not overkill ( noticeable quality improvement ).

Turning it all the way to 0 however for heaven well thats up for debate and i'm not going to be the one debating that as I'm well aware of how much gains are to be had.

Three things wrong with this conclusion:
Its not provable. This has been out since before Crysis 2 and the Tess design snafu using Tessellation. Its a benchmark, and every generation they have made it a little bit "tougher" on the cards. Even on the extreme setting running the benchmark, it only takes Tessellation to a scale of 1.0, the actual Tess slider inside Heaven goes to 2.0, which is extremely detrimental to performance. You claim that it favors nVidia, I could easily claim nVidia simply has better Tessellation hardware implementation.

Two its a benchmark designed to use Tessellation, Tessellation saw fit by the a 3rd party developer. Not by nVidia or ATi/AMD.

Another thing is that benchmarking competitively like this is just blatant cheating. If it was one or two points, it would be no different, but we are talking about upwards of 20~50% score increases over nVidia simply without tessellation. Artificially inflated scores of this magnitude are just cheating, flat out cheating.

If you guys don't want to fix that people can disable approximately 50% of this benchmark, I suggest you pull stop awarding points and go back to just displaying results. This benchmark has become entirely useless thanks to AMD.

P.S. In regards to LOD or Mipmap quality, there are some of us who think thats garbage too. Although its 2011, two wrongs still don't make a right.

[WhiteLion]

Vinster your is right and I agrees with you

[chew*]

Originally Posted by RagingCain

Three things wrong with this conclusion:
Its not provable.

Actually it is........

There is a cool program ( GPUperfstudio 2 ) you can use to see how much an object has been tesselated.

Best to run the program and see for yourself before spouting nonsense.

Here's an example of overkill tesselation.......if you cut this down to even 5% you would still not be able to spot any difference in quality visually.

[Massman]

As for the effect on the score

[El Gappo]

Wow, thought it would be much different from CCC to third party. How about 11? Is 2 points really worth the fuss?

Maybe we should be using game tests rather than these tessellation tech demo's.