Total War: WARHAMMER - AMD RadeonSI Linux Performance

[Azpegath]

We post our min specs on our store pages and usually via a news post as well prior to release. We find that making a clear post about what is an isn't supported is very useful for customers so they know what to expect and what they need to have a good experience.

I do have a correction to make, the article incorrectly states that the game is only supported when using:



we actually state you require:



It would be good to correct this mistake in the article as there is a large difference between recommending the latest official release of Mesa using the latest stable LLVM compiler and suggesting people have to use the latest git version of the drivers which are by their nature in development and change from one day to the next.

As active developers in the Mesa community since we started making games for the Linux platform we often use and contribute to the Mesa drivers so will use the latest git drivers to check out improvements that can help game performance or fix bugs but we will only officially support released driver versions not the latest git.

As Michael noticed there have been some improvements in the past week from Marek at AMD that show some positive signs in terms of improving performance in Total War: WARHAMMER and other titles. When the next stable release of Mesa is announced if these changes make a large improvement to the games performance we'll certainly recommend people upgrade.

Welcome to the forum, I'm glad to see a post from Feral (developers) themselves!

[Termy]

My R9 290 died (graphic corruption on screen, tested in 3 PCs with different specs) a couple months ago, and it suffered of the same performance issues as Michael's. Bridgman says they cannot replicate the problem, but is was as simple as installing Ubuntu 16.04 and upgrading the kernel to 4.7 using this repo:

http://kernel.ubuntu.com/~kernel-ppa/mainline/

Sadly, it's not that simple...
My R9 290 doesn't show this behaviour (at least it did not in kubuntu 16.04 with 4.7 and Manjaro with 4.6, 4.7 and 4.8)
I can believe that such a bug is a nightmare to pin down if not even the "same" hardware shows the same issues :/

[Michael]

Always take a note what Michael uses, most of the time it is something non default... he seems to tend to use modesetting ddx for all AMD hardware recently it seems

No I don't, I use the default.

[dungeon]

No I don't, I use the default.

Hm than it might be some kind of bug in suite, that reports modesetting most of the time

[pal666]

If the 290 has issues then you can bet your arse the 390 also has regression issues.

you can bet your arse that if 290 had issues they would be fixed long ago. some specific oem models of 290 have issues, but if developer has no access to it, he can't fix it.

[pal666]

Could someone here in the forums with the game and a 290/390 (non-X/X) give it a shot? I would greatly appreciate it!

you can do it yourself. if it does not work, you have 2 hours of playtime and 14 days of calendar time to return game for refund

[pal666]

Bridgman says they cannot replicate the problem, but is was as simple as installing Ubuntu 16.04 and upgrading the kernel to 4.7

no, it was not that simple. you missed step of shipping your card to them, because their cards do not fail

[Azpegath]

no, it was not that simple. you missed step of shipping your card to them, because their cards do not fail

Perhaps that would actually be an option. AMD would replace the card that exhibits the issue, with a replacement one or a new one (480 perhaps?). The user that sent in the card is happy with a new one, and all the other R290 users are happy that the bug is fixed.

[edddeduck_feral]

Could someone here in the forums with the game and a 290/390 (non-X/X) give it a shot? I would greatly appreciate it!

We've got one of those R9 290 (non X) within eye site of my current location and it runs the game just fine, it was used throughout WARHAMMER development in QA.

Right now we're running 16.10 with 13.1git as we're doing some regression/bug fix checks but we had it working fine with the latest official releases 12.1/13.0.1 as well. I don't know the brand but I think that model is an MSI card.

[bridgman]

If we can even find someone still seeing the problem who could provide more info that would help - the bug ticket activity has dried up since all of the recent reports of performance regression seem to have been things like having a dpm=0 setting left over from earlier experimentation. Will talk to the devs and find out what next steps should be... if Michael is still seeing the regression maybe we can work something out.