AMD Radeon RX 7900 XTX & RX 7900 XT Arrive For Linux Testing

[Michael]

Michael

Another possibility for a modern game benchmark to be added to your PTS has just appeared:

Dying Light 2

With the just released version 1.8, they've added a benchmark mode:



Again, I don't own this game because of the Denuvo DRM, but since Steam has a return policy within 2 hours of playtime, maybe giving it a try might be worth it, because the game itself already runs OOTB on Proton nowadays, according to ProtonDB.

In any case, good luck & have a nice Sunday!

Saw that the other day... Bought it, grepped through binaries, unfortunately no command line controls to initiate benchmark. Was hopeful since it has -nologos and various other launch options but alas no benchmark option. Steam refund already requested :/

[Linuxxx]

Michael

Another possibility for a modern game benchmark to be added to your PTS has just appeared:

Dying Light 2

With the just released version 1.8, they've added a benchmark mode:

In update 1.8 we would like to highlight a few features that were implemented along the quality of life changes in the game:

• Benchmark option on PC - you can start it from the video settings.

Again, I don't own this game because of the Denuvo DRM, but since Steam has a return policy within 2 hours of playtime, maybe giving it a try might be worth it, because the game itself already runs OOTB on Proton nowadays, according to ProtonDB.

In any case, good luck & have a nice Sunday!

[coder]

it's clear at this point that AMD is way behind on raytracing even without any dirty tricks from nvidia, and while i thought focusing on that was a little silly for the last 2 generations while most games didn't really do much with raytracing that wasn't gimmicky, it's starting to become more of an issue.

Agreed. I had also hoped to see more progress, but I might make the same decision in their shoes. It's a risky move for AMD to contend for the performance crown. They'll be fine, if all they do is offer a better value.

[smitty3268]

Maybe Portal RTX is tickling a bad case in AMD's driver? In that case, some driver optimizations could go a long way.

Portal RTX was written by NVidia, and as such it's almost certainly designed in such a way to take advantage of NVidia GPU features in a way that will be difficult for AMD to match. That's kind of the main point of releasing it, the same way that they designed things like NVidia HairWorks to perform poorly on amd cards back when they weren't very good at tessellation.

That said, it's clear at this point that AMD is way behind on raytracing even without any dirty tricks from nvidia, and while i thought focusing on that was a little silly for the last 2 generations while most games didn't really do much with raytracing that wasn't gimmicky, it's starting to become more of an issue. I still don't think it's at the point where it matters that much yet, but I'm not so sure it will remain that way over the next couple of years, and that starts to matter more now with the new generation of cards.

[Keats]

The outlook for RTRT on AMD looks even dire than I thought:

If their own numbers are confirming that they are unable to achieve atleast a 2× jump in performance without FSR, then what this means in practice is that Portal RTX will atmost deliver 2 FPS instead of 1 FPS @ 4K as is the case for the current top-end 6900 XT right now.

Truly remarkable...

Seeing as it's a title where AMD performance is clearly being crippled deliberately, I don't think Portal RTX numbers are particularly relevant.

[pete910]

But VRR is not working, right?

The GPU and TV always work at 120 Hz, even for movies or games that doesn't reach such a high framerate?

yes, vrr is

[coder]

The outlook for RTRT on AMD looks even dire than I thought:

If their own numbers are confirming that they are unable to achieve atleast a 2× jump in performance without FSR, then what this means in practice is that Portal RTX will atmost deliver 2 FPS instead of 1 FPS @ 4K as is the case for the current top-end 6900 XT right now.

Maybe Portal RTX is tickling a bad case in AMD's driver? In that case, some driver optimizations could go a long way.

In general, it's clear that AMD opted not to try and compete with Nvidia on ray tracing performance, which was probably a wise move. They wouldn't have beat Nvidia, and yet it would've come at the expense of considerable amounts of conventional rasterization performance or die area.

[Linuxxx]

From Anandtech's coverage of their launch event:

"AMD’s own performance slides tout anywhere between a 47% and 84% increase in RT performance.
Though it should be noted that AMD’s numbers are with FSR enabled;
so we cannot divorce these gains from any changes that improve FSR performance on the 7900 XTX."

OMG, I just noticed that diagram shifted the baseline of the new gen scores by using perspective and pushing them back in the frustum. Wow, that's really sketchy.

The outlook for RTRT on AMD looks even dire than I thought:

If their own numbers are confirming that they are unable to achieve atleast a 2× jump in performance without FSR, then what this means in practice is that Portal RTX will atmost deliver 2 FPS instead of 1 FPS @ 4K as is the case for the current top-end 6900 XT right now.

Truly remarkable...

[BraindeadBZH]

This cards look sick, can't wait to see how good (or not) the Linux support is to buy one.

[littlewu2508]

Yesterday I checked the ROCm-5.4 code base, and found RDNA3 already got supported in several math libraries like rocBLAS, rocFFF, etc. MIOpen not supported. Is it possible to run some ROCm benchmarks? @Michael