Using the Chrome OS Graphics Stack on Intel-Based Linux Desktops | Hacker News Reader

Using the Chrome OS Graphics Stack on Intel-Based Linux Desktops

01.org

124 pointsdoener23 comments

The title doesn't give justice to the content: it's about zero-copy texture upload in Chromium's Ozone-GBM when used on Intel integrated GPUs.

I'd be curious if having to upload textures is just a feature that's a hold-over from a history discrete desktop GPUs.

Videogame consoles have been doing "zero-copy" textures since they existed(except if you're on the PS3, but we won't go there). You also have things like EGL_image_external[1][2] that get you the same thing in OpenGL ES.

[1] https://www.khronos.org/registry/OpenGL/extensions/OES/OES_E...

[2] https://community.arm.com/graphics/b/blog/posts/eglimage---u...

Note that even if you share DRAM with the graphic chip, you may take an extra step anyway to rearrange the texture in a tiled format since it's more efficient for sampling.

If you really care about load time you can decode the image directly in a tiled layout in software but that's more annoying.

Yup, in games we used to call this "asset cooking" where you target the platform specific texture compression in addition to tiling that you get as a part of that format.

You get a nice 3-6x memory savings plus put less pressure on the texture samplers as memory bandwidth drops as well.

The article is dated two days ago.

Off topic: Wow, that domain name.. I wonder how long Intel has been holding on to that one? (No pun intended... Well, sorta).

Also off topic: I spy a Leopold FC660! Been interested but couldn't justify the price.

Totally worth it if you like Topre.

WHOIS says creation date 1998-01-15.

Reminds me of the short in The Animatrix, "The Second Renaissance"

http://matrix.wikia.com/wiki/01

Are there any performance benchmarks comparing ozone-gbm vs ozone-wayland ? It seems that the implementation that the article talks about is moot if the Wayland implementation has near equivalent performance..since the hardware layers are abstracted out.

In which case, this is only relevant for architecture that Wayland does not target. Most Intel desktops are supported by Wayland.

> Hardware overlay is not only for video, but also for rendering Android applications that run on a hardware overlay plane, such as the Wayland client of Chrome OS.

What?

What is a "Wayland client of Chrome OS" and how is it an Android application??

That sentence is quite confusing, to my knowledge a better explanation would be:

"The Android compatibility layer of Chrome OS implements a Wayland client to run on a hardware overlay plane."

based on information from https://www.phoronix.com/scan.php?page=news_item&px=XDC2016-...

Ah, so the Chrome OS UI, which is running directly on GBM/KMS, became a Wayland compositor just for the Android compatibility layer?

(Oh, and for the recently announced Linux desktop apps support maybe?)

The Wayland compositor was added to the Chrome OS UI to support Android, but it was very convenient for us to glob onto that for the Linux desktop apps support.

https://chromium.googlesource.com/chromiumos/docs/+/HEAD/con...

Awesome! I had played around with this code as well on Rockchip systems. Great to see the stack running on Intel.

How far did you get with Rockchip. I consistently ran into kernel/userspace driver version mismatches.

Google has all their code open, but the rockchip and MALI drivers must match Kernel version. Not much is out there, but hav you checker this?

https://github.com/rockchip-linux/libmali

Yes, those are the userspace drivers I was using. r1300 and r1400 are older than either the gru or mainstream kernels. It's a moving target, and rock chip hasn't been keeping up.

https://github.com/rockchip-linux/libmali/issues/20 was my ticket

How is this compare to Mozilla's Servo approach?

https://servo.org/

From the simple block diagram and benchmark, servo seems to be better?

Servo claims 500 fps.

This only shows 22-48 fps.

Anyone have good technical explanation on the difference?

You're comparing apples and tigers.

Intel's work is focused on the lower layers of the graphics stack, and removing the inefficiencies of older (X11 for example) architectures.

Servo is for rendering HTML, that is, taking HTML markup, and displaying the paragraphs, tables and whatnot graphically. So it is operating at a higher level.

"22-48 fps" is for video decoding

"500 fps" is for page rendering

Reply on news.ycombinator.com