Arch Linux User Repository

@ant0n et all, lets keep the comments here about packaging issues, general Brave usage issues should go in another forum to not clutter up this comment space. I'm deleting comments that have no relation to packaging. Grey areas like crashes that could be blamed on Arch can stay until proven otherwise, but things like how to configure Brave to handle popups or site X or whatever just don't belong here. Thanks for understanding.

@maxpayne3 Well if you want the details of my reasoning:

Linux is the kernel, yes, that doesn't matter; it's the potential for package-specific hacks that only concern itself with the userspace of Debian-derivatives that I'm concerned about. I don't want to maintain patches for those as a matter of principle, no matter how small, and even if there's none needed now, those packages are meant for Debian derivatives, meaning that upstream is free to break compat with us at any time for they don't officially support running the Debian package in a non-Debian system, and they very explicitly ship a tarball for non-Debian systems. Does this mean they'll add such hacks that break .deb's contents for Arch? Unlikely, but I'd rather stick to reasonable expectations for upstream: *.deb meant for Debian-based distros, Tarball is generic.

But even more important to me is the first thing I mentioned: I don't want to make a big packaging change for so small a gain, it's not worth my time. If your network constraints means that those 33MB are a big difference for you, I'd suggest forking the Google Chrome AUR and adapting it for Brave's deb release, and you can even publish it as a brave-deb AUR package for other people who share the same needs.

Hope that makes sense, sorry I didn't explain myself earlier. Cheers.

Sorry, but I don't think it's worth reworking the PKGBUILD and adding unnecessary risk using a release not meant for us for 33MB.

Not meant for us? Fedora and Ubuntu are Linux distributions, just like Arch. Those packages include even the desktop entry and logos, so you don't have to provide yours outside. Maybe it's worth give them a try. Google Chrome on AUR is made from the official deb package.

The scaling problem appears to have been fixed with the newest release. Thanks for the help again mixedCase (o:

@mixedCase Simply adding that flag does show video hardware acceleration on paper. But looking at any videos I found it's using ""Dav1dVideoDecoder" and looking at the CPU usage, I think it's not using GPU for video decoding.

I believe it used to have a build parameter that allows GPU video decoding. Can we get it back?

[update] Just checked the official deb packages. It seems they have switched to compile with vaapi off. I guess unless we compile it ourselves, there's nothing we can do.

[update] OK, filed an issue: https://github.com/brave/brave-browser/issues/16392

[update] My bad. I probably just played a video that's encoded using AV1. But most of other videos are indeed decoded using GPU. Confirmed by using the media tab of the inspector. Sorry for the confusion.

@mixedCase Thanks, that seems to work :)

Graphics Feature Status
Canvas: Hardware accelerated
Compositing: Hardware accelerated
Multiple Raster Threads: Enabled
Out-of-process Rasterization: Hardware accelerated
OpenGL: Enabled
Rasterization: Hardware accelerated on all pages
Skia Renderer: Enabled
Video Decode: Hardware accelerated
Vulkan: Disabled
WebGL: Hardware accelerated
WebGL2: Hardware accelerated

@maverick1 Sorry, but I don't think it's worth reworking the PKGBUILD and adding unnecessary risk using a release not meant for us for 33MB.

@francoism90 That feature moved to a command line parameter when the rebase to Chromium 91 occurred:

--enable-features=VaapiVideoDecoder

You can add it to your .config/brave-flags.conf for convenience.

Seems hardware video decoding support has been removed, is it possible to enable the video_decode flag again?

Can you switch to the .deb package instead of the .zip one? Its saves 33 MB of download.

@mixedCase Yup; running it with --force-device-scale-factor=2 fixes the problem for now.

@mixedCase Thanks; I'll have a look at it. And thank you for maintaining the package.