I had to just log in and say, Thank you for supporting the linux side of your game!
About stutterings, i noticed stuttering in Max Mad using Vulkan backend while using Gsync and OpenGL worked fine althou lower performance.
I noticed same effect in Doom 2016 (Wine). OpenGL was running OK, while Vulkan high FPS but massive frame lag, 200MS giving unplayable stuttering.
I’ll join in to say: Great job Haemimont Games, thanks for the Linux release and following up on it like this.
I just tried the Vulkan beta as well and the stuttering/performance rubberbanding issue I encountered is there as well. So whatever it is, it’s not tied to OpenGL. Feral know about it but can’t reproduce it so far. I e-mailed them asking if they could test on a Z77 and/or 3750K, maybe it’s something about that chipset/CPU combination that causes it.
Im using Nvidia 970 and AMD 9590 on 990FX with Kernel 4.10, so don’t think its chipset related.
These Mad Max performance regressions that you are reporting with both the OpenGL and Vulkan backend, do you have reason to believe that they are NVIDIA driver regressions? If so, why, and between what versions?
Thank you
I can’t say for sure that they’re related to Nvidia driver regressions anymore, since additional testing with Victor Vran revealed the performance drop in Victor Vran is not always on the same driver versions as the rubberbanding and stuttering in Mad Max. In the first week I had only tested some combinations where both games were showing issues on the same Nvidia driver versions.
It could be related to kernel 4.9 and newer, since the Mad Max issue does not appear on older kernels even when newer Nvidia drivers are used (e.g. 375.39 on 4.8 shows only slight stutter in the OpenGL version disappears in the Vulkan version).
Also, Tomb Raider is showing consistently same-ish performance on all configurations. So it might be an issue in the Mad Max engine.
Are there other games I should try? I am looking at Divinity: Original Sin, Enhanced Edition now.
I tested some other games now and tried to go for a big variety of engines and technologies to cover all sorts of cases. The spreadsheet’s updated. There were no further issues in any other games, apart from microstutter in Portal 2, but that is a known problem for Source games.
The Mad Max issue remains, but at this point I’m almost sure it’s an engine issue and not a driver one.
So it’s pure coincidence that I happened to play the only two titles that exhibit bad behavior at the same time and precisely after Debian rolled out a new driver. That exposed the Victor Vran issue which I previously hadn’t encountered, and the Mad Max issue had been there all along.
I will go through the rest of the titles on Ubuntu with kernel 4.8 as well just to be complete, but I expect everything will be in order. Thanks again for looking into this, I’m happy that at least Victor Vran should get his old performance back for some more vampire hunting, and I hope that the Mad Max issue sorts itself out.
Psy-Q, try Mad Max with Steam set to offline mode and see if it solves the issue. Works for me.
We have investigated this issue and found this is not nvidia driver issue. Lets close this thread. If you have issue with other games then its good to open new thread to avoid mess.
Wow, this is interesting! I just tried this, and performance is fine. I don’t know why online mode only works well with older drivers/kernels but this is a solution while Feral investigates further.
Sorry to have suspected an Nvidia driver issue, but two coincidences combined to make me guess that after my first tests: The new Nvidia driver version with fixed behavior for glBufferStorage exposed an issue in Victor Vran, and perhaps kernel 4.9’s network behavior (?) is exposing an issue in Mad Max.
The rest of the games are not affected so I’m happy to close this. Thank you everyone involved for your time and effort.