AMD Squeezes In Some Final AMDGPU Changes To DRM-Next For Linux 4.21

[bearoso]

The cursor plane fast path is not to be underestimated. Chromium scrolling goes from choppy to butter-smooth in all cases, and a bunch of small, mysterious, hiccups everywhere just go away.

Thanks, Nicholas and driver team!

[Big_Mama]

edit: forget it

[agd5f]

There's one open source AMD driver for recent chips, amdgpu. There are a number of ways to get it however. It's upstream so you'll end up with it as the in box driver in most distros. We also provide driver packages (same code) for supported distros for users that want to use a packaged driver rather than the in box driver. It's ultimately the same driver.

[Big_Mama]

But they do have, at least for now(?), different features? And different performance? And it doesn't even look like one is just an earlier version of the other because both have their own strengths and weaknesses. Well you have "X.Org ATI Driver Developer" written below your username, so I guess you know what you are talking about but to me that doesn't look so clear. And I bet I'm not the only one

[clapbr]

The cursor plane fast path is not to be underestimated. Chromium scrolling goes from choppy to butter-smooth in all cases, and a bunch of small, mysterious, hiccups everywhere just go away.

Thanks, Nicholas and driver team!

this x9000

we had to insist a bit too much on bugzilla but Nicholas saved the day

I have been using the first patch on my 4.19 ever since he posted it and it is night and day for a lot of my daily uses. Things that were annoyingly choppy became butter-smooth as said above. It is unfortunate it isn't planned to be backported to 4.19 LTS.

[fuzz]

But they do have, at least for now(?), different features? And different performance? And it doesn't even look like one is just an earlier version of the other because both have their own strengths and weaknesses. Well you have "X.Org ATI Driver Developer" written below your username, so I guess you know what you are talking about but to me that doesn't look so clear. And I bet I'm not the only one

The kernel driver is the same. The opengl/vulkan implementations differ.

FYI, this is the person you responded to https://www.phoronix.com/scan.php?pa...q=Alex+Deucher

[agd5f]

But they do have, at least for now(?), different features? And different performance? And it doesn't even look like one is just an earlier version of the other because both have their own strengths and weaknesses. Well you have "X.Org ATI Driver Developer" written below your username, so I guess you know what you are talking about but to me that doesn't look so clear. And I bet I'm not the only one

Same features. Like I said, it's the same code. That said, the code is constantly evolving with new features and new asic support being added, so the upstream kernel 4.18 may have less features for a particular asic compared to kernel 4.19. Same with packaged releases. With the exception of the packaged releases, it's no different from any other open source driver.

[euler271]

this x9000

we had to insist a bit too much on bugzilla but Nicholas saved the day

I have been using the first patch on my 4.19 ever since he posted it and it is night and day for a lot of my daily uses. Things that were annoyingly choppy became butter-smooth as said above. It is unfortunate it isn't planned to be backported to 4.19 LTS.

do you happen to know that for a fact? because, if there's no explicit statement saying that's the case, I think you might only need to send a request to the mailing list for the stable kernel and make your case. at the very least you'll get the explicit answer. I did that a couple of days ago and ask them to commit a revert patch for ia32 that was crashing my kernel during boot since around 4.17, and in a day or so they added it to 4.19.9.
edit: ofc, in my case, it was just a single commit involving only one file, I dont know how complex yours is

[clapbr]

I did test open source freesync with the Rocket League game: https://www.youtube.com/watch?v=lGesNjhUo-w

I hope they merge those patches to Mesa git asap.

See the artifact at 00:15, I had those occasionally plus some white flashing.