ocaholic - Board

Mit der EAH 6950 präsentiert ASUS seine erste AMD Radeon HD 6950 basierte Grafikkarte. Diese kann auf insgesamt 2 GByte GDDR5 Grafikspeicher zurückgreifen kann. Der Nachfolger der Radeon HD 5850 lässt auf einen ordentlichen Performance-Sprung hoffen, der nicht nur durch höhere Taktraten realisiert wurde.



ASUS EAH 6950 2GByte

Jau, das mit der Performance habe ich mir eben auch gedacht. Irgendwie werd ich einfach nicht wirklich warm bei der Karte. Sie ist ja toll und gamen kann man damit prima ... aber eben ... irgendwie weder Fisch noch Vogel das Ding ...

Welcome on board my friend.

Criticism is always welcome! But it stops at the point when someone starts blaming me for things I'd absolutely never do. I don't think that you've read many of our reviews because if you did so you would have noticed that we verbalize what we've tested. When a product is good then it's good and we explain why. When it's bad then it's bad and we explain why.
As an example the following conclusion:
http://www.ocaholic.ch/xoops/html/module…el_lang=english

Furthermore, sure it's true, we do not have DX11 game benchmarks in our comparison tables. We do test 3DMark 11 which - as you surely know - supports DX11.

Regarding the Core i7 at 2.66 GHZ and game settings at 8x AA 16x AF: Have you ever done scaling tests?
Take for example the following article: http://www.legionhardware.com/articles_p…_hd_5970,3.html

For Batman Arkham Asylum you see a 9 Percent performance increase with a 100 Percent clock increase --> the CPU is definitely not limiting anything here
For Wolfenstein you see a 3 Percent performance increase with a 100 Percent clock increase --> again the CPU is not limiting at all
For Left4Dead you see a 15 Percent performance increase with a 100 Percent clock increase --> between 2.0 GHz and 2.4 GHz you see that there is a jump afterwards, almost no difference
For Modern Warfare 2 you see a 14 Percent performance increase with a 100 Percent clock increase --> again between 2.0 GHz and 2.4 GHz there is a jump afterwards almost constant
For Company of Heroes you see a 15 Percent performance increase with a 100 Percent clock increase --> again between 2.0 GHz and 2.4 GHz there is a jump afterwards almost constant
It goes on like this for Crysis Warhead, Far Cry 2, BattleForge and H.A.W.X.

Worring about CPU scaling would start when the 100 percent clock increase would show a 100 percent performance increase in the game. Differences between 0 and 15 Percent are insignificant regarding a 100 Percent CPU clock increase.

Another thing is that Legion Hardware did some test with no AA/AF or low AA/AF settings. This means less load on the GPU, thus the likelyhood that the CPU could be limiting increases.



My 50 cents to the 6950: I would actually buy it. I like the card and it's well priced but facing the truth that its architecture has been overworked the performance increase noe gets isn't significant enough. This is my opinion.

I have already been thinking about adding the Unigine benchmark to the selction and I think it will be a good addition. In this case I'll bench as many cards as possible with this benchmark too.

In our graphics benchmarks you'll never find benchmarks for which we used fraps, because the measurement error is simply too big. What we do use are benchmarks which come with a built in routine such as Resident Evil 5, World in Conflict, ... have one. But at this point we could also start a debate about theoretical and practical benchmarks and this one wont end. Our approach is to use time demos or theoretical benchmarks.

Personally I think Alien vs. Predator is not a very good benchmark. It doesn't use that many dx11 effects and therefore it would be just one benchmark more.
Regarding Stalker, does this have an integrated benchmark routine or would you have to use fraps? If fraps there is no sense in doing this. The same for Dirt 2. If it has an integrated routine which runs automatically then it's worth considering, otherwise it does not make much sense.

The Stone Benchmark also looks good imo but Dirt 2 I wouldn't use.