NVIDIA vs. AMD 2D Performance Benchmarks

[entropy]

Not using, but tried it now... works tad fast here wit Catalyst . xterm too .

I remeber that with xterm on radeon driver too, but that slowness some people hit actually depends on font setup... with some fonts both drivers can be slow there .

Thanks. I'll have a look into the font setup.

[mmstick]

try with performance governer

http://openbenchmarking.org/result/1404050-PL-1404055PL56

hd 7790
OpenGL
Text = 303\213
Gears = 473\250
Image Scaling = 5259\2720

XRender
Text = 181\91
Gears = 161\85
Image Scaling = 331\173

Here's my results:

I decided I didn't have anything better to do at the moment, so I benchmarked my Radeon HD 7950 on both Catalyst and the latest Gallium3D/Mesa stack with Kernel 3.14.4 on KDE 4.13, with compositing disabled, using QGears2. Results are sorted fastest to slowest:

With performance governor:

Gallium3D/Mesa Git + Kernel 3.14.4 - OpenGL
Text = 458.68 FPS
Gears = 420.62 FPS
Image Scaling = 4114.69

Catalyst 14.4v2 + Kernel 3.14.4 - OpenGL
Text = 449.80
Gears = 412.27 FPS
Image Scaling = 4230.50 FPS

Gallium3D/Mesa Git + Kernel 3.14.4 - XRender
Text = 148.67 FPS
Gears = 131.78 FPS
Image Scaling = 881.90 FPS

Catalyst 14.4v2 + Kernel 3.14.4 - XRender
Text = 144.92 FPS
Gears = 130.23 FPS
Image Scaling = 823.08 FPS

Without performance governor:

Catalyst 14.4v2 + Kernel 3.14.4 - OpenGL
Text = 429.46 FPS
Gears = 396.20 FPS
Image Scaling = 3562.73 FPS

Gallium3D/Mesa Git + Kernel 3.14.4 - OpenGL
Text = 403.97 FPS
Gears = 366.57 FPS
Image Scaling = 2099.33 FPS

Catalyst 14.4v2 + Kernel 3.14.4 - XRender
Text = 141.54 FPS
Gears = 125.73 FPS
Image Scaling = 636.55 FPS

Gallium3D/Mesa Git + Kernel 3.14.4 - XRender
Text = 138.88 FPS
Gears = 122.50 FPS
Image Scaling = 620.69 FPS

[Pontostroy]

Here's my results:

I decided I didn't have anything better to do at the moment, so I benchmarked my Radeon HD 7950 on both Catalyst and the latest Gallium3D/Mesa stack with Kernel 3.14.4 on KDE 4.13, with compositing disabled, using QGears2.

Great work. Radeonsi beat catalyst. What cpu you have?

[mmstick]

Great work. Radeonsi beat catalyst. What cpu you have?

In this desktop, a 3.8 GHz FX-8120.

[toguro123]

Throw a Matrox graphics card in this test just to get a little bit of historical perspective...

[zanny]

I'm glad 2D is going away. These benchmarks just show how AMD probably invested quite a bit of effort into 2d over 3d and that doesn't really matter anymore.

Once we are all on Wayland anyway, and all the GPU drivers are using GLAMOR in the xwayland DDX for just backwards compatibility (at least for the sane foss drivers, heres hoping Catalyst is just left behind entirely and Nvidia can suffer a bit providing their own xwayland driver layer) and egl / opengl for everything in the native desktops, there won't even be any 2D anymore.

[md1032]

Do the Nvidia results improve at all if you change the performance governor from adaptive to max?

[dungeon]

Do the Nvidia results improve at all if you change the performance governor from adaptive to max?

Those are power states, but i can be wrong for nvidia . I see radeon driver also have those changible, so maybe there are also some difference between balanced and performance but those differences should be seen only on laptops i think .

http://www.botchco.com/agd5f/?p=57

[molecule-eye]

Firefox SMOOTH scrolling performance is butter with Gallium and MUCH slower using Catalyst 14.4. That's weird given these benchmarks. I wonder if it's a problem particular to Firefox. Anyone else seeing this?

[entropy]

Firefox SMOOTH scrolling performance is butter with Gallium and MUCH slower using Catalyst 14.4. That's weird given these benchmarks. I wonder if it's a problem particular to Firefox. Anyone else seeing this?

Do you run a compositor?
From my experience 2d performance significantly drops for Catalyst + Compositor.