Chrome 80 Against Firefox 74/75 Performance On Linux

[timrichardson]

Eh, I've been using it for a while on OpenSUSE Tumbleweed (where the chromium is patched to be able to use HW accel) and it's like 25% of the videos and many livestreams are psychedelic colors and light shows.

Wish you better luck than me.

On KDE Neon (ubuntu LTS) I'm using this build with good success. Thinkpad t480: https://launchpad.net/~saiarcot895/+.../chromium-beta
some of the official distribution builds are very careful/strict about the formats they decode due to licencing issues. That PPA is less encumbered by these considerations. The dev even explains how to get Widewine DRM working.

[Bestia]

Thanks, are you using an Intel processor with an iGPU or is the RX 5600 xt the only GPU in the system? Afaik the hardware acceleration is still available in Intel iGPUs even if you install an additional card.

No I have Ryzen 5 3600.

[starshipeleven]

You're barking up the wrong tree. It's a chromium bug, same as or similar to https://gitlab.freedesktop.org/gstre...api/issues/136 . Other drivers are just hiding it by not exposing 10 bits per component configs (which breaks use-cases where 10 bits per component are required or at least desirable).

No it isn't a Chromium bug, after I got reports that all is fine on everything I did some more testing.

There is something wrong in my current installation, as running a newly-installed OpenSUSE Tumbleweed on the same hardware does not have that problem.
Since I do tinker with stuff I'm probably at least partly responsible for this I guess.
Just deleting and reinstalling Chromium does not fix the issue so I'm probably going to migrate my stuff to a new install.

[MrCooper]

Surprise surprise, I just tried them on my work laptop with an Intel CPU/graphics, and its playing fine with hw accel enabled.

Goddamnit AMD.

You're barking up the wrong tree. It's a chromium bug, same as or similar to https://gitlab.freedesktop.org/gstre...api/issues/136 . Other drivers are just hiding it by not exposing 10 bits per component configs (which breaks use-cases where 10 bits per component are required or at least desirable).

[starshipeleven]

I have zero issues with W10 right effing now.

*pirated Enterprise long time supported version

It doesn't spy on me either, since I'm running W10 LTSC.

You can't prove this.

I have made the following argument, "W10 spying has not been proven by anyone yet"

and you refuse to acknowledge that what was written in the license agreement by highly payed lawyers isn't proof enough.

[pgoetz]

Not caring about their performance difference enough to switch.
Firefox does a good enough job for me. I'd rather take the small performance hit than get more Google infestations than I already have.

^---- This. I'm losing the ability to tell where google ends and I begin.

[birdie]

And while you're hating me for speaking the truth and providing hard evidence, I help resolve critical issues. Fan-atics are so re tarded unfortunately. Imaginary privacy W10 issues make it worse than perpetually half-baked, half-broken, half-complete desktop Linux.

[birdie]

I have zero issues with W10 right effing now. It doesn't spy on me either, since I'm running W10 LTSC. Linux right effing now has TONs of issues. Linux fan-atics just cannot stop making up BS, LMAO.

I have made the following argument, "W10 spying has not been proven by anyone yet" (I'm not talking about LTSC specifically, I'm talking about any W10 version). Instead of actually giving me a counterargument, I'm shown some vague M$ document which the person didn't even read. LMAO.

Your argumentation sucks because it's void and nil. God, people are insane in their bigotry.

[R41N3R]

KDE user? Then it's a KDE bug:

https://bugzilla.mozilla.org/show_bug.cgi?id=1567762

https://bugs.kde.org/show_bug.cgi?id=411682

I'm a KDE user. But as Firefox is the only application with copy and paste issues, I'm not confident to state it is a KDE bug. There is one in Gnome too:

https://bugzilla.mozilla.org/show_bug.cgi?id=1622538

[archsway]

Afaik in ChromeOS the "browser" interfaces with the kernel modules directly to access hardware acceleration (also on ARM-based Chromebooks for that matter), there is no VAAPI middleman protocol/libraries or anything of the sort. Having full control over the whole firmware kind of helps there.

Actually, video acceleration for Chrome on CrOS goes through libv4l2, at least for most of the ARM chromebooks and possibly a few Intel ones.

There are a couple of VPU drivers in the mainline kernel with a v4l interface, but not for Intel/AMD.