5040x1050 Performance report for all you 5760x1200 sufferers

[kondor999]

After implementing the brilliant hack detailed here - http://www.widescreengamingforum.com/forum/viewtopic.php?f=64&t=17635&start=0 - I am now, at last , able to do 5040x1050 and get a real feel for the performance difference from 5760x1200.

Here's a preview: It's huge. I will update this post as I test new games, but I wanted to go ahead and post an important one:

Dirt2 (results copied directly from DiRT2_Benchmark)

5760x1200 2xAA DX9 (you have to edit hardware_settings_config to get this, and it's well worth it since DX11 adds nothing and sucks performance in this game) Preset "High":

average min_fps="27.544434" av_fps="40.070484"

5040x1050 2xAA Everything else same:

average min_fps="45.572693" av_fps="58.110287

Notice that minimum FPS. That is just so huge in a driving game - or anything else, for that matter. Much better experience.

Note: This is with Crossfired 5870's. See my sig for the rest.

[skipclarke]

I'd like to see what you get in DIRT2 in DX10. All the testing i did with Far Cry 2 showed big performance gains using DX10.

[Dave Baumann]

Small point - DiRT 2 doesn't hav a DX10 mode, its DX9 or DX11 only. :)

[kondor999]

Small point - DiRT 2 doesn't hav a DX10 mode, its DX9 or DX11 only. :)

That's correct. Here's something to illustrate what I mean re: the performance hit that DX11 represents:


As you can see, the 5970 drops from 113 to 72 fps. And, according to this article http://www.pcgameshardware.com/aid,699998/Dirt-2-DirectX-9-vs-DirectX-11-graphics-compared-Update-New-comparisons/Practice/ (and my own observations), it's virtually impossible to see the difference between DX9 and DX11 while actually playing.

[fatlazyhomer]

1650x1080 wasnt supported?

[Dave Baumann]

As you can see, the 5970 drops from 113 to 72 fps. And, according to this article http://www.pcgameshardware.com/aid,699998/Dirt-2-DirectX-9-vs-DirectX-11-graphics-compared-Update-New-comparisons/Practice/ (and my own observations), it's virtually impossible to see the difference between DX9 and DX11 while actually playing.

AFAIK the DX9 mode of DiRT2 does the post processing a 1/4 the display resoultion, resulting in a more blurred image in saome cases; under DX11 the pos processing is done at full resultion, using high qualify FP16 formats, via ComputeShader. There are cases where this i more noticable in use; I still think the lighting in the night stadium races look great in DX11.

[kondor999]

[quote]As you can see, the 5970 drops from 113 to 72 fps. And, according to this article http://www.pcgameshardware.com/aid,699998/Dirt-2-DirectX-9-vs-DirectX-11-graphics-compared-Update-New-comparisons/Practice/ (and my own observations), it's virtually impossible to see the difference between DX9 and DX11 while actually playing.

AFAIK the DX9 mode of DiRT2 does the post processing a 1/4 the display resoultion, resulting in a more blurred image in saome cases; under DX11 the pos processing is done at full resultion, using high qualify FP16 formats, via ComputeShader. There are cases where this i more noticable in use; I still think the lighting in the night stadium races look great in DX11.
Yep, you're correct again. DX9 looks blurrier - which I happen to think looks different, but not necessarily worse. Still, the DX11 mode is clearly doing more and that's why it's slower.

Dave - you are a damn fount of wisdom, you know that? I may decide to actually listen to you :wink:

[Abram]

I expected a larger difference between the minimums...
Makes me a feel a bit better about going for 16:10 monitors and gaming at 5760x1200, instead of 16:9 displays and 5760x1080..
Hey abou tbenchies for those two? or x1080, anyway

[skipclarke]

Small point - DiRT 2 doesn't hav a DX10 mode, its DX9 or DX11 only. :)

That's kind of a shame. Only option is a small (at least to some people) improvement for a big fps hit. While a DX10 mode would have offered a (possibly) small improvement, while giving a sizable fps improvement.

I expected a larger difference between the minimums...
Makes me a feel a bit better about going for 16:10 monitors and gaming at 5760x1200, instead of 16:9 displays and 5760x1080..
Hey about benchies for those two? or x1080, anyway

The slightly wider aspect of 3x16:9 might actually give you a lower performance.

[kondor999]

I expected a larger difference between the minimums...

Believe me, the difference between 27 and 45 min fps is night and day in terms of actual gameplay. The former isn't even playable - the latter imperceptible.