Quoting: damarrinShmerl, let's not continue this fairly pointless back and forth. I don't think I can say anything that'll convince you, or that you can say anything that'll convince me. We're all Linux brothers, regardless of the CPU, GPU or distro we use.

You're not really helping your case with claims like "AMD don’t let people play games on Linux". Please don't go around starting fires if you don't like the flames.

AMD's driver situation for their new hardware is far from perfect (on Windows as well I hear), but that obviously doesn't mean that AMD doesn't let us play our games on Linux.

I have a perfectly reasonable response to that, but I'm just going to stop posting now. Do note what thread we are in, though.

Quoting: damarrinI have a perfectly reasonable response to that, but I'm just going to stop posting now. Do note what thread we are in, though.

Yeah, and I get the frustration. I'd be frustrated as well if I bought hardware and it didn't work.

But my point was that your attitude seems very "us versus them" for someone who thinks we're all brothers. I wasn't really trying to start a conversation.

Quoting: Shmerl

Quoting: barottoIt turns out to be a sensors related problem. Apparently you can't use any utility that periodically polls the sensors or the desktop freezes with powerplay messages sent to syslog.

Yes, this bug has been an issue from the start, but it got somewhat better. Some fixes are still pending. If it will be still a problem in 5.5, it means something is still missing.

The curious thing is I went along using my desktop with just the nodma trick enabled, and everything was apparently fine. Then powerplay problems started after I made a system update, even if I locked my kernel version to v5.3.0-19, which gave me no problems before. This is when I tried v5.4.1, which did not resolve anything.
Maybe navi firmware files got updated? Does that make sense? Is it possible the problems are in the firmware rather than in the kernel? Or maybe it's both...

Quoting: jensI guess bottom line is:
- Go AMD for a perfectly integrated Linux solution, but either wait a year after release of a new hardware generation or be prepared to invest considerable time into living on the edge with kernel RC's, mesa master etc.
- Go Nvidia if you want to use newest hardware but be prepared to restrict your use cases to what NVidia supports, thus e.g. just classic X and be ready for the occasional breaking when updating your system.

Both options have their own advantages, non is perfect, so best answer here: it depends on personal preference ;)

I think this sums it up quite nicely.

Quoting: jensI guess bottom line is:
- Go AMD for a perfectly integrated Linux solution, but either wait a year after release of a new hardware generation or be prepared to invest considerable time into living on the edge with kernel RC's, mesa master etc.
- Go Nvidia if you want to use newest hardware but be prepared to restrict your use cases to what NVidia supports, thus e.g. just classic X and be ready for the occasional breaking when updating your system.

Pretty much, yes. I expect AMD to improve that, once their financial situation will improve as well. For instance, they can start providing upstream support more in advance, like Intel do, hire more developers for the Linux team to fix bugs faster, and so on.

However that's not the only criteria for the choice, at least for me. Nvidia not working properly with upstream due to stubborn refusal to open source and problems they cause to Nouveau is a big minus. And in contrast, reporting upstream issues to AMD and Mesa is a big plus for both users and developers. Not only AMD can work on those, but the likes of Valve and others can participate, thus we get radv, ACO and so on, which wouldn't have been even possible otherwise.

Last edited by Shmerl on 1 December 2019 at 6:13 pm UTC

Quoting: barottoThe curious thing is I went along using my desktop with just the nodma trick enabled, and everything was apparently fine. Then powerplay problems started after I made a system update, even if I locked my kernel version to v5.3.0-19, which gave me no problems before. This is when I tried v5.4.1, which did not resolve anything.
Maybe navi firmware files got updated? Does that make sense? Is it possible the problems are in the firmware rather than in the kernel? Or maybe it's both...

That's interesting, but if I understood it correctly, there was a race condition in powerplay, which was fixed in 5.4. So not sure why the problem still persisted for you with 5.4.1. For me it got fixed, but those messages still spam dmesg, so I disabled most amdgpu sensors meanwhile, except for temperature monitor in KDE plasmoid, which doesn't trigger it for me.

Last edited by Shmerl on 1 December 2019 at 7:00 pm UTC

Quoting: damarrinShmerl, let's not continue this fairly pointless back and forth. I don't think I can say anything that'll convince you, or that you can say anything that'll convince me. We're all Linux brothers, regardless of the CPU, GPU or distro we use. From things you've written in the past I can see you're a sensible chap, I'll let you live in your all-is-dandy-in-AMD-land insanity bubble if you let me live in my la-la-la-nvidia-la-la-la insanity bubble, how about that?

The point is not to bubble anything, but to explain objective issues. No solution is perfect.

WOw wish I knew this before buying an AMD for the first time in 15 years, I only read how awsome AMD now was in linux for the last few months. And ofc I bought a 5700 xt thinking since its been out for a while now the drivers would be rock solid. But now I read that they pretty mutch suck.

For a month or so it's actually pretty good, but you need to use cutting edge stack for it. And it's generally well known, that for using Mesa and AMD you are better off with rolling distros. If that's news for you, then good to learn it now :)

Last edited by Shmerl on 3 December 2019 at 4:26 pm UTC