ODROID-N2 was announced last February for $63 (2GB RAM), and $79 (4GB RAM), but Hardkernel was not quite ready to take orders at the time. One of the good news is that the 4GB RAM is now available for pre-order with shipping scheduled to start on April 3.
Another good news is on the software side with Hardkernel having released the userland Mali-G52 Wayland driver. It does not work well with Linux 4.9 due to incomplete DRM implementation, but it goes work with Linux 5.0 plus some modifications as further discussed in the aforelinked forum thread.
The screenshot above, courtesy of odroid forum member memeka , shows ODROID-N2 running Ubuntu 18.04 + Gnome3 + Linux 5.0 on top of Wayland with GPU drivers providing acceleration as shown by glmark2-es2-wayland test program.
The benchmark results are pretty impressive:
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 |
======================================================= glmark2 2014.03+git20150611.fa71af2d ======================================================= OpenGL Information GL_VENDOR: ARM GL_RENDERER: Mali-G52 GL_VERSION: OpenGL ES 3.2 v1.r16p0-01rel0.2943fc4ef9657d91ee32c9a58dec6cd2 ======================================================= [build] use-vbo=false: FPS: 961 FrameTime: 1.041 ms [build] use-vbo=true: FPS: 1592 FrameTime: 0.628 ms [texture] texture-filter=nearest: FPS: 1491 FrameTime: 0.671 ms [texture] texture-filter=linear: FPS: 1477 FrameTime: 0.677 ms [texture] texture-filter=mipmap: FPS: 1524 FrameTime: 0.656 ms [shading] shading=gouraud: FPS: 1151 FrameTime: 0.869 ms [shading] shading=blinn-phong-inf: FPS: 1215 FrameTime: 0.823 ms [shading] shading=phong: FPS: 1043 FrameTime: 0.959 ms [shading] shading=cel: FPS: 1126 FrameTime: 0.888 ms [bump] bump-render=high-poly: FPS: 514 FrameTime: 1.946 ms [bump] bump-render=normals: FPS: 1976 FrameTime: 0.506 ms [bump] bump-render=height: FPS: 1777 FrameTime: 0.563 ms [effect2d] kernel=0,1,0;1,-4,1;0,1,0;: FPS: 1139 FrameTime: 0.878 ms [effect2d] kernel=1,1,1,1,1;1,1,1,1,1;1,1,1,1,1;: FPS: 383 FrameTime: 2.611 ms [pulsar] light=false:quads=5:texture=false: FPS: 2096 FrameTime: 0.477 ms [desktop] blur-radius=5:effect=blur:passes=1:separable=true:windows=4: FPS: 389 FrameTime: 2.571 ms [desktop] effect=shadow:windows=4: FPS: 788 FrameTime: 1.269 ms [buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 103 FrameTime: 9.709 ms [buffer] columns=200:interleave=false:update-dispersion=0.9:update-fraction=0.5:update-method=subdata: FPS: 129 FrameTime: 7.752 ms [buffer] columns=200:interleave=true:update-dispersion=0.9:update-fraction=0.5:update-method=map: FPS: 158 FrameTime: 6.329 ms [ideas] speed=duration: FPS: 356 FrameTime: 2.809 ms [jellyfish] <default>: FPS: 979 FrameTime: 1.021 ms [terrain] <default>: FPS: 52 FrameTime: 19.231 ms [shadow] <default>: FPS: 437 FrameTime: 2.288 ms [refract] <default>: FPS: 88 FrameTime: 11.364 ms [conditionals] fragment-steps=0:vertex-steps=0: FPS: 1769 FrameTime: 0.565 ms [conditionals] fragment-steps=5:vertex-steps=0: FPS: 1769 FrameTime: 0.565 ms [conditionals] fragment-steps=0:vertex-steps=5: FPS: 1853 FrameTime: 0.540 ms [function] fragment-complexity=low:fragment-steps=5: FPS: 1783 FrameTime: 0.561 ms [function] fragment-complexity=medium:fragment-steps=5: FPS: 1726 FrameTime: 0.579 ms [loop] fragment-loop=false:fragment-steps=5:vertex-steps=5: FPS: 1715 FrameTime: 0.583 ms [loop] fragment-steps=5:fragment-uniform=false:vertex-steps=5: FPS: 1650 FrameTime: 0.606 ms [loop] fragment-steps=5:fragment-uniform=true:vertex-steps=5: FPS: 1728 FrameTime: 0.579 ms ======================================================= glmark2 Score: 1119 ======================================================= |
I’ve never seen such as high score (1,119 points) on Arm hardware. But at the same time, I’ve never run the wayland version of the benchmark, so that may partially explain it. For reference, RockPro64 RK3399 SBC gets 48 points in the onscreen test (default), and 201 points in the offscreen test. I could not find results for glmark2-es2-wayland for Rockchip RK3399 despite being supported. So we’ll have to see what part of the much higher score is part of the Mali-G52MP GPU, and which part is the use of Wayland over Xorg/X11.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress
As far as I remember from the N1, glmark2-es2-wayland results were in the 800-900 range for RK3399, but with a 32bit blob and userland. But I doubt it would have been bigger on 64bit userland.
Wayland, yay! Good job, as usual! : )
> glmark2-es2-wayland results were in the 800-900 range for RK3399
And how is it possible that Jean-Luc only got a score of 48 points when testing? Is this just the difference between X11 and Wayland?
Two words: buffer management. For the on-screen tests, the presentation part of the pipeline has been the bane of most GLX/EGL stacks. It doesn’t matter how quick a GPU is, if the presentation part does memcpy between FBs/RTs/pixelbuffers/etc. For the off-screen tests, one’d need to profile, but I’d wager some buffer management shenanigans at play as well. Wayland traditionally solves the presentation part.
For the onscreen tests, I think the maximum framerate is also limited by vsync. For example if output if 1080p50, the benchmarks may not go over 50 fps.
That’s right, vsync is a factor in on-screen. I was referring to the registered difference in performance of N1 under wayland and x11 whether on-screen or off-screen.
I half recall reading some where that the drivers only work on the VPU, is that right?
We’re talking about the GPU drivers here. i.e. for 3D graphics acceleration.
The VPU is for video encoding/decoding.
I’d suggest taking Linux 5.x benchmarks with a pinch of salt, as apparently a lot of hardware benchmark a lot higher on Linux 5.x, but there’s no real world performance gain.
For real? I saw some Linux 5.0 benchmarks on Phoronix and there was a decrease in performance in some tests
Apparently on some ARM hardware at least. Not sure about other platforms.
what about 4k opengles?
glmark should support something like offscreen rendering. so it should be possible to run test with 4k resolution
To date, a large majority of MALI models, can support Wayland. I am just as curious, to see turning Kodi with a VPU able to be used, something that I doubt following my many tests.
If the Amlogic codecs are still usable on the Kernel 5.0, we will have to use Framebuffer, or Xorg.
there are no x11 drivers, arm stopped providing them in recent mali DDKs.
for N2, there are no amlogic codecs on 5.0, but they are working for the older amlogic socs (s805, s905). so i’m sure there will be soon for N2 too.
Kodi is ok on wayland on gbm.
I did not know for “X11”, remains to see if an buildroot has AMLogic is planned officially, with the use of the Framebuffer and these codecs.
I doubt the Dream team is designing their own video accelerator for the “Dreambox One”, I think, they relied on the ports (/dev/amstream, /dev/amvideo…), and official codecs.
Very bad radiator design, board will overheat during benchmark and any CPU intensive task.
BS
LOL. Don’t be ridiculous.
I assume this is another joke on you 🙂