Superposition Shows How Far RadeonSI Gallium3D Has Evolved vs. AMDGPU-PRO

[phoronix]

Phoronix: Superposition Shows How Far RadeonSI Gallium3D Has Evolved vs. AMDGPU-PRO

Comparing the hybrid AMDGPU-PRO proprietary Linux driver to the RadeonSI Gallium3D open-source driver stack with the newly-released OpenGL 4.5-using Unigine Superposition has shown how far the open-source driver stack has come...

http://www.phoronix.com/scan.php?page=news_item&px=Superpos-PRO-RadeonSI

[eydee]

It's not just about how far RadeonSI advanced. AMDGPU-PRO still has remnants of the terrible Catalyst drivers in its OpenGL implementation. It doesn't have optimal OpenGL performance. This is why Vulkan tests would be more interesting. Only Unigine decided it's not mature enough... F... them.

[bug77]

It's not just about how far RadeonSI advanced. AMDGPU-PRO still has remnants of the terrible Catalyst drivers in its OpenGL implementation. It doesn't have optimal OpenGL performance. This is why Vulkan tests would be more interesting. Only Unigine decided it's not mature enough... F... them.

Even so, it's an indicator. Just look at the performance a year or two ago.
The real issue as I see it, is OpenGL applications require profiles to reach optimum performance and AMDGPU will probably never get those profiles. Vulkan on the other hand shifts that burden onto developers. How that's going to work out, we don't know yet.

[Dr. Righteous]

It's not just about how far RadeonSI advanced. AMDGPU-PRO still has remnants of the terrible Catalyst drivers in its OpenGL implementation. It doesn't have optimal OpenGL performance. This is why Vulkan tests would be more interesting. Only Unigine decided it's not mature enough... F... them.

Really the only linux benches that catch my eye are linux vs windows on the same systems. I want to know how well the vendors are supporting linux with drivers.
I can never expect open source driver to preform as well as ones provided by the manufactures. And in the case they out preform the OEM drivers, it shows the lack of effort on their part to provide good drivers.

[duby229]

Really the only linux benches that catch my eye are linux vs windows on the same systems. I want to know how well the vendors are supporting linux with drivers.
I can never expect open source driver to preform as well as ones provided by the manufactures.

But you can always expect them to be a hell of a lot more stable and reliable.

If what you want is maximum performance where the graphics does not reflect what the game designer intended, then sure go for proprietary drivers. It may not look like the game devs intended but hey, your getting 300FPS on your 60hhz monitor, cool huh!

[bug77]

But you can always expect them to be a hell of a lot more stable and reliable.

If what you want is maximum performance where the graphics does not reflect what the game designer intended, then sure go for proprietary drivers. It may not look like the game devs intended but hey, your getting 300FPS on your 60hhz monitor, cool huh!

I don't know about that, I've been running Unigine benchmarks on both Windows and Linux, I haven't seen any differences in rendering. "does not reflect what the game designer intended" seems to be implying the proprietary drivers are rendering different things altogether which I believe we can all agree is not the case.

[imirkin]

Michael, in case you want to do a similar NVIDIA blob vs nouveau comparison with a selection of Kepler/GM107 GPUs, you have to run superposition with MESA_GL_VERSION_OVERRIDE=4.5 MESA_GLSL_VERSION_OVERRIDE=450, as nouveau does not presently advertise GL 4.5 and it seems like Superposition requires it (or maybe it requires 4.4, I didn't test that). It did seem to run fine on my GT 730 (GK208) board... like 10-15fps at 1024x768 and all settings on low.

[duby229]

I don't know about that, I've been running Unigine benchmarks on both Windows and Linux, I haven't seen any differences in rendering. "does not reflect what the game designer intended" seems to be implying the proprietary drivers are rendering different things altogether which I believe we can all agree is not the case.

Absolutely it is the case. Yes most definitely.

[babai]

Really the only linux benches that catch my eye are linux vs windows on the same systems. I want to know how well the vendors are supporting linux with drivers.
I can never expect open source driver to preform as well as ones provided by the manufactures. And in the case they out preform the OEM drivers, it shows the lack of effort on their part to provide good drivers.

Hey! AMD employs many RadeonSI OSS driver developers! Which is exactly why you see boost in performance in the last two years.

[andre30correia]

again?! why AMD continues with two official drivers? Why they don't simply give up from closed one? and put more ppl working with open ones?