OpenGL vs. Vulkan On The AMD Ryzen 3

[phoronix]

Phoronix: OpenGL vs. Vulkan On The AMD Ryzen 3

We have previously looked at Vulkan vs. OpenGL Linux game CPU core scaling and Linux game scaling across multiple CPUs but at the time did not have a Ryzen 3 system. Now having Ryzen 3 Linux box, here is a look at how the Vulkan versus OpenGL performance compares on the low-end processor. As well, it's a fresh look at the NVIDIA vs. RadeonSI/RADV performance.

http://www.phoronix.com/vr.php?view=25067

[raonlinux]

Interesting benchmarks, btw is possible with a 480 or 580 have mesa opengl and amdpgu pro vulkan installed at the same time?? I m been looking for some info but I hadn't luck. I will to try opengl with mesa and vulkan with the pro version for dota 2.
Greetings.

[GruenSein]

Something is really odd here. Iirc, the latest RadeonSI benchmarks showed an RX580 generally performing on par with the GTX1060 with more tests in favor of the RX580. The RX480 tested here is almost the same card. It is not clocked quite as high but this shouldn't cause this much of a difference. Either Mesa requires a lot more CPU cycles than the Nvidia driver which really shows on this low end CPU or something else is weird. Anyway, the difference seems crazy large...

[bug77]

Something is really odd here. Iirc, the latest RadeonSI benchmarks showed an RX580 generally performing on par with the GTX1060 with more tests in favor of the RX580. The RX480 tested here is almost the same card. It is not clocked quite as high but this shouldn't cause this much of a difference. Either Mesa requires a lot more CPU cycles than the Nvidia driver which really shows on this low end CPU or something else is weird. Anyway, the difference seems crazy large...

Yup. I own a GTX1060 and I've read a lot of benchmarks before I bought it. Never saw that big a difference.

[cj.wijtmans]

Something is really odd here. Iirc, the latest RadeonSI benchmarks showed an RX580 generally performing on par with the GTX1060 with more tests in favor of the RX580. The RX480 tested here is almost the same card. It is not clocked quite as high but this shouldn't cause this much of a difference. Either Mesa requires a lot more CPU cycles than the Nvidia driver which really shows on this low end CPU or something else is weird. Anyway, the difference seems crazy large...

can you link some benchmark to disprove micheal?

[schmidtbag]

Kind of funny, the industry is trying to slowly move away from quad cores for gaming, and yet with Vulkan all it seems to do is breathe more life into quad cores. Not that I'm complaining - that just means we get to pay less on hardware.

[VikingGe]

Interesting benchmarks, btw is possible with a 480 or 580 have mesa opengl and amdpgu pro vulkan installed at the same time??

In theory, yes (check this out) - but be warned, I tried that a few weeks ago and it kind of broke both libGL and Vulkan, leaving me with a working KDE desktop and playable Doom, but glxgears and anything using Vulkan that isn't Doom would complain about missing drivers.

[bug77]

can you link some benchmark to disprove micheal?

Nobody's trying to disprove Michael. We're just noting that either there's a regression in there or some peculiarity with the setup. Those cards are usually within 10% of each other.

[cb88]

http://www.phoronix.com/scan.php?pag...0xt-vega&num=2

Yeah, the RX480 scored twice what the 1060 did in this run of Dota 2 in that test at the same resolution... and Ryzen 3 is still a reasonably fast single threaded CPU so that doesn't' really explain the difference.

The RX480 score is like 4-5x higher in that test...

In general the opensource Radeon driver should be alot faster than the Nvidia blob except for the very high end cards ie a RX480 still won't beat a 1080 or 1080ti.... despite having a better driver.

[VikingGe]

One thing I noticed on my RX 480 is that DPM doesn't seem to ramp up clock speeds high enough when a game is primarily bottlenecked by the CPU:
power_dpm_force_performance_level = auto
power_dpm_force_performance_level = high

Granted, it's only a ~20-30% difference, but it might contribute to the poor radeonsi performance in this particular test.