2016 End-of-Year Open-Source Radeon Benchmarks With Linux 4.9, Mesa 13.1-dev On Many Different GPUs

[phoronix]

Phoronix: 2016 End-of-Year Open-Source Radeon Benchmarks With Linux 4.9, Mesa 13.1-dev On Many Different GPUs

With 2016 soon drawing to an end, it's time for all of my year-end recaps now of Linux drivers that I have been doing for the past 12 years. Today are benchmarks of a wide assortment of AMD graphics cards on both R600g and RadeonSI Gallium3D drivers when using Mesa 13.1-dev + LLVM 4.0 SVN and the Linux 4.9 kernel for providing a bleeding-edge look at the open-source AMD Linux graphics performance across hardware going from the Radeon HD 4890 series all the way up through the RX 480 and R9 Fury hardware. Here's a fun look at the OpenGL driver performance across this range of GPUs.

http://www.phoronix.com/vr.php?view=23882

[SaucyJack]

That whole not being able to hit at least 120 or 144 fps in some games is really annoying.

[smitty3268]

Anyone running the r600g driver should apply this patch: https://lists.freedesktop.org/archiv...er/135975.html

If fixes a bunch of lockups in various games. Not sure if it would fix Michael's 4850/4870 problems or if that's something else.

Unfortunately nobody has reviewed/committed it to master, because everyone's busy looking at radeonsi now.

[Jumbotron]

Looks to me that these tests show clearly two things.....

The best bang for the buck overall is the RX 470. But if 2 GB of RAM is all you need and getting a card on Ebay for less than $150 is more important then the R9 285 is the best bang for the buck.

[Creak]

I thought the RX 480 was better than the RX 470, but in these tests it's just slightly better, and the 470 is better than 480 in quite a few benchmarks. How so?

[Michael]

I thought the RX 480 was better than the RX 470, but in these tests it's just slightly better, and the 470 is better than 480 in quite a few benchmarks. How so?

Its a factory overtclocked RX 470.

[Creak]

Its a factory overtclocked RX 470.

Badly overclocked apparently...

[ElectricPrism]

I thought the RX 480 was better than the RX 470, but in these tests it's just slightly better, and the 470 is better than 480 in quite a few benchmarks. How so?

"Better" is relative. For example in Team Fortress 2 the test is CPU-bound and not GPU-bound, therefor a lot of cards render a similar framerate.

There are also other factors too suchas I'm sure 8GB of address space on RX 480 has some overhead however small vs the 4GB on the RX 470 in the case of the TF2 test showing RX 480 performed something like 0.3 FPS less.

Then I would imagine too that the MESA driver has a role too and game optimizations, while it may be possible technically for RX 480 to outclass RX 470 I imagine that the current driver and games have room for optimizations to increase that gap - which is a matter of time & work IMO.

So there's my ideas on what other factors could play a role in your observation.

[Creak]

You're right ElectricPrism, especially for the CPU bound part where neither the driver, nor the GPU is responsible. For the rest, let's say that for the 50 extra bucks to get an RX 480 isn't really worth it.

(BTW I have an RX 480 and I'm very happy with it, but the facts are that I would probably not advise an RX 480 over an RX470 since the benefits aren't really obvious -- Edit: yet)

[mibo]

Thank you Michael for this mountain of tests.
It also shows how much the drivers were improved - really impressive.