Total War: WARHAMMER III Linux Performance Across 24 NVIDIA / AMD GPUs

[phoronix]

Phoronix: Total War: WARHAMMER III Linux Performance Across 24 NVIDIA / AMD GPUs

Last week Feral Interactive released Total War: WARHAMMER III for Linux as a native port of this game that debuted earlier this year on Windows. This latest Total War game port to Linux by Feral relies on the Vulkan API for graphics and comes with a modest system requirements. For those curious about the performance of this latest high profile game port, here are benchmarks across 24 different AMD Radeon and NVIDIA GeForce graphics cards on Ubuntu Linux.

https://www.phoronix.com/vr.php?view=31229

[bosjc]

Good benchmark. Would be really interesting to look at the Feral port vs. Proton, too, across NVIDIA and AMD GPUs. I know at least on my setup, I saw a pretty good performance boost on the port vs proton.

[birdie]

I'm more curious whether Michael could replicate this https://fedoramagazine.org/fedora-wo...ing-far-cry-5/

I suspect something was terribly wrong only I can't say what exactly.

[obri]

Good benchmark. Would be really interesting to look at the Feral port vs. Proton, too, across NVIDIA and AMD GPUs. I know at least on my setup, I saw a pretty good performance boost on the port vs proton.

Woud be really interesting for me too AS I want to play it with friends that use Windows.
So I have to use proton.

Michael Could you do some tests with Proton as comparision to the port?

[Volta]

I'm more curious whether Michael could replicate this https://fedoramagazine.org/fedora-wo...ing-far-cry-5/

I suspect something was terribly wrong only I can't say what exactly.

Everything is as it should be. To be honest it looks like a memory leak in drivers on Windows.

[piotrj3]

I'm more curious whether Michael could replicate this https://fedoramagazine.org/fedora-wo...ing-far-cry-5/

I suspect something was terribly wrong only I can't say what exactly.

I would expect that some intensive graphical feature on linux is missing and is only implemented as stub that does nothing.

[user1]

I'm more curious whether Michael could replicate this https://fedoramagazine.org/fedora-wo...ing-far-cry-5/

I suspect something was terribly wrong only I can't say what exactly.

Yeah, I've seen this. I wonder why he didn't specify exactly the hardware that was benchmarked. He said nothing other than "RDNA 2 based GPU", which sounds like he's hiding something.
I mean it's true that in some cases AMD gpu's may perform a bit better on Linux in some games, but these results indeed make me think there's something wrong with his setup.

[spykes]

I would expect that some intensive graphical feature on linux is missing and is only implemented as stub that does nothing.

Maybe or it's simply a windows driver bug/ bad implementation (it happens as well).

[birdie]

I would expect that some intensive graphical feature on linux is missing and is only implemented as stub that does nothing.

Yeah, I've asked for PNG screenshots but I don't think the author is serious about this whole thing. "Linux runs faster than Windows! Migrate now!" that's it.

[Jabberwocky]

Nvidia is running away in the power consumption sections with highend cards averaging around 270W!

The RX 6800 is looking attractive for a long term Linux gaming card. Perf/Watt looks superb in these tests... it's the same as a RTX 3060.

I'm more curious whether Michael could replicate this https://fedoramagazine.org/fedora-wo...ing-far-cry-5/

I suspect something was terribly wrong only I can't say what exactly.

Probably just a DXVK bug related to that specific game or hardware.

If all games showed a "62.52% margin as compared to those on Windows 10 Pro 21H2" while keeping the same graphical fidelity then everyone and their dog would be installing Linux and every tech YouTube channel would be talking about it. I don't have any problems with someone investigating strange stats and blogging about it without getting to the root cause. I have a problem with someone blogging about half researched things and making strange conclusions like "Performance improvements like these only make me want to indulge more in testing and finding out what else Fedora Workstation is capable of" ... I mean heck just go to read the Inferences section. That article is around the level of The Verge's PC Build Video. I'm hoping someone reaches out to the author because people reading that **** is going to be very confused and it could be damaging to the community.