3D performance on the resolution and antialiasing filter

[Jaabili]

Here are the results of the resolution and frame rates for anti-aliasing (AA) separately itemized. This facilitates the comparison, if one is looking for a graphics card for a certain screen size or resolution.

Hoping that my test would have somebody over there.

[Abigail101]

Test is great. Missing single HD5870 and HD5770 in CF. And it would not be wrong in one or two sentences to write. But really Nice test. Since I have benched including AvP also have two questions:

1. If the FPS Average and maximum detail? The values of the GTX in DX11 mode are not really comparable, because the fallback and the lack of a tessellation can stand the Nvidia cards a little better than they are.
   
2. And second: I found that CF scales at just AvP not really very good if the tessellation comes into play, then break the FPS namely a sometimes even to levels below those of a single card. The phenomenon I had the way for other DX11 games as well as Metro. Have you noticed anything like this? I wanted to put a second 5870 in my PC, but I am purely commercial in many games and applications in the red.

I find the results of the CF-small teams quite remarkable, but unfortunately a lot of DVDs on the shelf, where games are nice, can do with CF at all. Then the value of such a vehicle combination is even below zero. I think you should prepare the user mentally and morally to the fact that this maximum results are probably more the exception however. Even in 2D a CF team is made up of two 5870, still behind a single card. Filed, no profile, no summation of the performance. Is like SLI, only gets jerky.

[Umberto-Micro]

Yes, they are really strong, it also has its advantages when the 3D engines are a little longer on the market, as the driver optimizations are faster there. The 5750er work with the CPU together really well. 5670 I have planned for the next test, the rest comes with the following tests. Powercolor was informed that react very slowly, maybe it works the next Test. Micro stuttering, should be tested specifically and more.

[Kanakpriya]

The benchmark suite they scale super, 85% of what you want. In the last suite, games were there that ATI does not really wanted because it was constantly bitching. In this suite, everything is going good, now fit again not, you can never make right, I'm not the games, even the patches, and the driver. No, not noticing that it would also be seen in the results. This behavior is typical of Metro for some games forging, which was also for 3 months with Stalker, then up to 8 patches and 3 drivers, memory management with Crossfire works reasonably and in sync running. Two ways to make single cards or another game has been optimized to that. As has been seen in The Last Remnant or Hawx, can also no longer be repaired. For me, that are no longer a reason against Crossfire or SLI, support for 2 years is okay in the popular titles after 2-3 months. A major factor is the driver support after 2 years, if only optimized for new generations and the still relatively strong team is sorted out ugly. Currently, the lifetime for X2 and CF 2-3 years, is slowly thereafter ugly.

[Osman84]

I agree to fully Kanakpriya there. When you off in the media game poaching regions then single card is definitely preferable. Although it appears even the games do not run or graphic errors. It is after all connected with immense costs expensive programmers to put on a driver profile, because it does not pay for the very last game (which will definitely never be used in the benchmarks) to sacrifice resources that can be seen in the form of tickets sold again. That's a economic decision and nothing else. For the games has become a great support. You see here for example in Battlefield Bad Company 2. For me the 2nd map scales well and is studying for the FPS to 60 at maximum details and filters. Also 35 in single card provides funds. These values I find here again if I compare the GTX260 216SP and the GTX295.

[Efigenio]

That will always exist, the differences between the technologies are just there. Nvidia also has advantages with hardware PhysX, where ATI has nothing. This was already found on ATI and improved AF filtering, a common denominator in all accessible technologies in 3D graphics and impossible just window dressing. Different quality of filtering, other basic position on the drivers, paid support games in development, different optimization for SLI and Crossfire, custom 3D engines for a manufacturer, PhysX, different graphics memory, HDR-R DX9 ... There are three options:

• All technologies into separate charts
• All to a common denominator
• All comparison

[Arumugan]

Splitting does not help because it compares only one manufacturer is saying that a 5770 faster than a HD 5750 is obtained, without saying, the performance difference would not matter, because the distance is only a pure number without reference to the competition would be. (The first thing I would do at separate tables, I was watching, where the competition lies with the frame rate, then compare the look and weigh the difference in performance with the graphics advantage. The intermediate step can be omitted in AVP, graphic differences for only certain things (Aliens) hard to see, and rarely, so the performance difference is almost insignificant.) Bring everything to a common denominator, would not take advantage of new technologies. I want to know how much difference is this. Many people had long been a X800 because of connecting would have meant to DX10 fewer frames for more money. There's still many games DX9 or reuse (Assassins Creed 2), one can see no visual benefit.

[Jaabili]

AVP is so minimal in the performance difference that I have decided not to make separate charts. Would Metro 2033 in the suite, where DX11 is widely used, there would naturally separate charts for DX10 and DX11 with all cards for HD5000er and 400. This will also come at some point. In AVP, which uses DX11 tessellation and very sparingly, not necessary because the difference, as described in the text, is at 1-3 fps, which is actually measuring tolerance, each driver update can cause the 5-fold change. To make the work of the computer screen bad, to justify an argument is, but not necessary. The wording is a bit contentious, but the question was still okay. Your opinion is no one you finish, you've tried it with the alleged lack of information, I have shown where they stand in the article and copied it to you an excerpt. Then the incompatible, the case for a common chart I have described in detail, it is up to you how you deal with it. It can also be positive, you will not be the only one who sees through the charts and the text of which only flies over. This is not new or unusual, but we have the forum and I will answer factual questions.