Dolphin, the GameCube and Wii emulator - Forums

Core i7 980X @ 4.45 ghz - GTX 580 @ stock

windowed 1024 x 768

Normal
1x 306 fps
2x 216 fps
3x 150 fps

4 SSAA
1x 213 fps
2x 105 fps
3x 59 fps

9 SSAA
1x 145 fps
2x 57 fps
3x 27 fps

Edit: accurate data

Clusta, did you changed the default resolution in the test build? You're supposed to leave it at 640x480 (fullscreen display resolution) and 1024x768 (windows size). The test build settings should not be changed beside the efb scale (1x-3x) which are allowed to changed. Any who, those are great results for 1920x1200 resolution with the given specs.

If he changes the screen resolution it won't change the internal resolution, framerate will be identical or nearly identical.

Look at the gtx 460 and 580 comparison (the first two on the graph), the gtx 580 lost in 1x-3x efb scale and you still saying the res don't matter? -_-

(01-22-2011, 11:09 AM)tuanming Wrote: [ -> ]Look at the gtx 460 and 580 comparison (the first two on the graph), the gtx 580 lost in 1x-3x efb scale and you still saying the res don't matter? -_-

Tested on mine, did all my results in the 1024x768 window, setting my full screen res to 1680x1050 and going full screen gave the same results. The resolution that the wii is rendering at depends on the efb scale.

The window/fullscreen resolution only matters if using fractional or integral scaling.

Anyway, my results:

xeon 3520 (i7 920) @ 2.8ghz/HT, 6gb ram 1333, gtx 275

Normal
1x 187 fps
2x 151 fps
3x 110fps

4 SSAA
1x 151 fps
2x 75 fps
3x 43 fps

9 SSAA
1x 106 fps
2x 41 fps
3x 20 fps

xeon 3520 (i7 920) @ 4.2ghz/HT , 6gb ram 1600, gtx 275

Normal
1x 282 fps
2x 169 fps
3x 117 fps

4 SSAA
1x 170 fps
2x 79 fps
3x 44 fps

9 SSAA
1x 113 fps
2x 41 fps
3x 20 fps

Quote:Look at the gtx 460 and 580 comparison (the first two on the graph), the gtx 580 lost in 1x-3x efb scale and you still saying the res don't matter? -_-

Just look at the source code. The only time the backbuffer resolution is even mentioned is in the scaled copy from efb to backbuffer.

@ninjin
My point is that the high-end gtx 580 on the higher clocked 980x should not lose to a mid-range card on a lowered overclock 970x...

This is what you should do for your test: Just clock your cpu to 4.2Ghz and use the default settings on dolphin 640x480 (fullscreen display resolution) and 1024x768 (windows size) with 1x-3x efb scale, then compared it to this configuration--1680x1050 (fullscreen res) and 1680x1050 (windows size). You should NOT get the same results with 2 different configurations, and that's my point.

Clearly that guy did not followed the rules.

For those who are curious to know the results at the default fullscreen res, I have posted them. There isn't a huge difference, esp. at high end settings where the people's interest lie.

Thanks for the updates Clusta. Based on your updated data, I can safely say that the gtx 580 is an overkill for dolphin. However, the gtx 460 still beat your gtx 580 at 2x efb scale, what the heck is going on here? Got the latest driver? Were you doing any background work while doing the benchmark?

He probably has the NV contral panel settings set to high quality while the 460 is at default settings though it shouldn't make that big of a difference.