Radeon Vulkan Driver Added To Mesa, Fresh Radeon Vulkan vs. OpenGL Benchmarks + AMDGPU-PRO

[smitty3268]

You're probably thinking of Windows d3d vs Linux ogl. If I remember right, Windows ogl vs Linux ogl are roughly the same.

For non-Skylake.

[boltronics]

Nice thing I noticed was that (for the Fury at least), free software Vulkan was beating free software OpenGL. Since I was using Mesa with OpenGL previously anyway, that's already a win.

[dungeon]

Aside from the Vulkan results, the other exciting take-away was seeing how RadeonSI Gallium3D's OpenGL performance is now outperforming the AMDGPU-PRO OpenGL driver!

To be exact, that is 10% more fps on Dota, but 40% slower in Talos. That is in my book... 30% slower on average

[dungeon]

I would be excited if i see shader cache and threaded GL in opensource driver.

It peak in some case, while being half slow in many other ... same we can say about red vs green proprietary comparison, so it is boring to me as diff is the same as last year or even 12 years ago

[chimpy]

For non-Skylake.

For non-Skylake? If by that you mean every other integrated Intel graphics and all GPUs created by nVidia and AMD then; thats cool.

[Veto]

To be exact, that is 10% more fps on Dota, but 40% slower in Talos. That is in my book... 30% slower on average

You might want to find another book on arithmetic: (10-40)/2 = -15. So 15% slower on average. But the validity of averaging percentages is another discussion...

[Adarion]

That's looking quite promising already. Indeed, AMD will have to come up rather quickly with their contributions.
But then, according to John Bridgman, they're currently quite busy with the preparations for the upcoming HW.

In other news I wonder who on earth (besides Michael) uses a 3840 x 2160 resolution for gaming. Are there 8k gamers out there? Multi-monitor-gamers? My max is still 1920 x 1200 and below (and the 1200 is because I love 16:10 or 4:3, some people also gotta do work on their boxes).

[dungeon]

You might want to find another book on arithmetic: (10-40)/2 = -15. So 15% slower on average. But the validity of averaging percentages is another discussion...

Yeah different books tell us different stories... everybody is free to diminish an issue and move on

[CrystalGamma]

You might want to find another book on arithmetic: (10-40)/2 = -15. So 15% slower on average. But the validity of averaging percentages is another discussion...

AFAIK you usually use geometric mean on percentage differences, so e^((ln 1.1 + ln 0.6)/2) = 0.81240384, so about 19% slower … ☺

[dungeon]

But in case of RX 480 - 35 vs 57 fps, i would say big diff there

I can also pretend that i don't know some things about it, also that next pro will have faster GL profile there... and in ideal second run (one shader is cached) and with multithread disabled i can easely predict 65 fps there...

That way difference became even bigger, but OK i will shut up about unreleased things for now