Radeon Linux Gaming Performance: Ubuntu 17.04 vs. Ubuntu 17.10

[GreatEmerald]

Michael makes sure to mention where there's a bug with CS:GO/Gnome where there's basically a popup from Gnome asking if the application is frozen when CS:GO launches. When a normal user is playing, they'll just click the "Wait" button in the dialog, and then everything will be fine. Unfortunately, I suspect that the automated benchmark isn't very good at this :P

As such, the CS:GO benchmarks should be ignored for their X results as they aren't indicative of actual gameplay. Which is kinda the point of benchmarks. To, y'know, tell you how the game is gonna perform when you play it.

I wasn't talking about CS:GO. I meant that (X)Wayland is slightly faster than Xorg in most of the cases.

[theghost]

Nice how in many cases the Xorg performance is lower than that of (X)Wayland, I didn't expect that.

It looks so on paper but if you have to use input it will most probably hit: https://bugzilla.gnome.org/show_bug.cgi?id=745032
Another gem: https://bugzilla.gnome.org/show_bug.cgi?id=788493

[Brisse]

It looks so on paper but if you have to use input it will most probably hit: https://bugzilla.gnome.org/show_bug.cgi?id=745032
Another gem: https://bugzilla.gnome.org/show_bug.cgi?id=788493

Exactly. Real life gaming is still better on Xorg, despite what the benchmarks says.

[Azpegath]

I'm missing the R290.. Perhaps that would be a more relevant card to test than the Fury. Didn't they sell like 200 of the Fury cards?

[Brisse]

I'm missing the R290.. Perhaps that would be a more relevant card to test than the Fury. Didn't they sell like 200 of the Fury cards?

200 seems a bit off, but they are indeed rare. I feel lucky to be sitting on one of the Sapphire Nitro R9 Fury's. It's been a great card to me, and unlike my previous Asus ROG Matrix R9 290, I've been very happy with the Fury. I appreciate the inclusion of the Fury in all these articles

[Brisse]

Yea these xwayland benchmarks are not indicative of real performance. I just tried some CS:GO, and despite having the FPS counter stay at 200+ fps, it stuttered like crazy.

Not to mention the input lag. It's acceptable in most games but not something like CS:GO.

phoronix Would you consider not posting more of these xwayland benchmarks until these issues are solved? They are kind of pointless and misleading in my opinion. At the very least, the benchmarks should be accompanied by a description of subjective experience. It's one thing to have an automated benchmark running with no input, and it's an entirely different thing to actually play the games.

[pal666]

Will people switch to the the nouveau driver?

smart people will switch from nvidia hardware

[DarkFoss]

Because most Ubuntu users probably aren't enabling PPAs or the like for getting newer Mesa releases?

Trying to install either the oibaf or padoka ppa's on 17.10 is a mess more so padoka because of the difference in files. Ubuntu patched libdrm-amdgpu1 to depend on libdrm-common. Lot's of fun messages like:

dpkg: error processing archive /var/cache/apt/archives/libdrm-amdgpu1_2.4.83+git1709271040.7c7118~gd~a_i386.deb (--unpack):
 trying to overwrite '/usr/share/libdrm/amdgpu.ids', which is also in package libdrm-common 2.4.82-1

Leaving you with having to run apt --fix-broken install 1 or more times depending on which ppa you are trying to install possibly dpkg --purge libdrm-common if you upgraded from Zesty to Artful. I can only guess that enabling the dependency is part of getting AMD's ROCm & Vulkan working natively in the future? Removing libdrm-common for now doesn't seem to matter much but if it's related to DC it could lead to an interesting choice for users until it's sorted out in the ppa's .

On a happy note once sorted out padoka's ppa fixes the Steam code guard errors(vulkan only) I've been seeing with Serious Sam Fusion using oibaf's ppa.