Bug 86413

Summary: Euro Truck Simulator 2: Severe stuttering with Kernel >=3.17
Product: Mesa Reporter: Andreas Grois <dragon.arises>
Component: Drivers/Gallium/radeonsiAssignee: Default DRI bug account <dri-devel>
Status: RESOLVED FIXED QA Contact:
Severity: normal    
Priority: medium    
Version: 10.3   
Hardware: x86-64 (AMD64)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments: Screenshot of HUD on a bad configuration
Screenshot of HUD on a good configuration (Mesa 10.3 + Linux 3.16)
Screenshot of HUD on a bad configuration (Mesa 10.3.2 + Linux 3.17.2)
journalctl output while running ets2 on a bad configuration
dmesg of the bad configuration
Screenshot of HUD with drm-next-3.19 and Mesa from git

Description Andreas Grois 2014-11-17 21:51:21 UTC
Created attachment 109641 [details]
Screenshot of HUD on a bad configuration

When running Euro Truck Simulator 2 with the combination of Mesa 10.3 and Linux 3.17 or newer, the game shows severe stuttering whenever a new object enters the field of view (gets loaded to GPU RAM?) or the camera is rotated. The game stops for a short period of time, up to several seconds, just to return to a normal framerate afterwards. This stuttering is not present when combining either Mesa 10.2 with Linux 3.17, or Mesa 10.3 with Linux 3.16. With Mesa 10.4 from git the stuttering is present on recent Linux versions, but much less severe than with Mesa 10.3.

This issue might be related to bug:
https://bugs.freedesktop.org/show_bug.cgi?id=84570

Hardware: 
GPU: Radeon R9 270X (vendor overclocked)
Mainboard: Asus M5A88V-EVO (PCIe 2)

Software: 
xorg-server 3.16.99.901
xf86-video-ati-7.5.0
Mesa 10.3.3 or Mesa 10.2.9
Kernel 3.16.5, 3.17.2, 3.18-rc5
radeon-ucode-20140823
Comment 1 Andreas Grois 2014-11-17 21:52:14 UTC
Created attachment 109642 [details]
Screenshot of HUD on a good configuration (Mesa 10.3 + Linux 3.16)
Comment 2 Andreas Grois 2014-11-17 21:53:26 UTC
Created attachment 109643 [details]
Screenshot of HUD on a bad configuration (Mesa 10.3.2 + Linux 3.17.2)

Resubmit due to wrong file type of previous upload
Comment 3 Andreas Grois 2014-11-17 21:54:31 UTC
Created attachment 109644 [details]
journalctl output while running ets2 on a bad configuration

Not that I think that this contains something useful, but attached is the journalctl output that was recorded while the game was running on a bad configuration.
Comment 4 Andreas Grois 2014-11-17 21:55:11 UTC
Created attachment 109645 [details]
dmesg of the bad configuration

Again, probably not very helpful, but here's the dmesg output of a bad configuration.
Comment 5 Michel Dänzer 2014-11-18 02:02:50 UTC
Is it still bad with a kernel from Alex Deucher's drm-next-3.19 tree and Mesa Git master? If so, can you attach a HUD screenshot corresponding to that?
Comment 6 Andreas Grois 2014-11-18 18:39:32 UTC
Created attachment 109679 [details]
Screenshot of HUD with drm-next-3.19 and Mesa from git

I've tested the latest git version of mesa combined with a drm-next-3.19 kernel, and indeed the issue seems to be fixed with this combination.
There are some very minor stutters still, but those are hardly noticable and I'm not certain if they are related to the bug, or caused by something else (disk activity,...).
Comment 7 Michel Dänzer 2014-11-19 01:37:52 UTC
Excellent, so can this report be resolved?
Comment 8 Andreas Grois 2014-11-20 22:21:53 UTC
In my opinion: Yes. I'll mark it as resolved/fixed.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.