Was doing some performance testing with Basemark today. Most interesting factoid was the performance hit for redirected windows. On Vulkan this was only around 2%, but OpenGL took a hit of over 10%.

Clean results:




Memory was not detected, but it is 32GB @ 3200 MHz.

i7-3770 @ 3.4GHz (Stock)
16GB 1866MHz Patriot RAM
MSI GeForce GTX 1650 Super Gaming X 4GB

Mine is also displaying 0 ram at 0 mhz, I already tried sudo and running it as root. Also my scores aren't showing up on the website.

Any ideas?

Is there a trick to getting scores submitted and showing?

Last edited by Dragunov on 28 April 2020 at 6:57 pm UTC

Quoting: The_AquabatI recommend to not take this benchmark seriously on AMD, and it's not a driver problem, please don't blame driver developers, on SuperPosition benchmark the performance of this 5600xt is more or less what I expected somewhere near a vanilla 1080 or a 1070ti, little less than a rtx 2060.

I just installed the benchmark from flatpak and ran the official test, didn't bother taking shots because I don't really care about benchmarks as long as all my games run fine. I got a score of around 6500 with OpenGL and just slightly under 8200 on Vulkan with my 5700XT / Ryzen 7 3700X. I suppose the OpenGL test might be hitting some slow paths on Mesa, but a few percent faster than a 1080 Ti on Vulkan seems about right based on the reviews I saw back when I bought the GPU.

Last edited by tuubi on 21 August 2020 at 7:25 pm UTC

Ran again

Vulkan: 8280


OpenGL: 10520


Ryzen 3700X/GTX 1080Ti

Last edited by Xpander on 21 August 2020 at 8:01 pm UTC

Quoting: The_Aquabat

Quoting: tuubiI got a score of around 6500 with OpenGL

well on my system I'm getting about 1500 read my previous posts, slower than a rx570. This is with a 5600 xt.
why is so difficult to admit that the test is bugged on AMD?

Hey take it easy. I didn't say it was working fine. The OpenGL test obviously doesn't give comparable scores for AMD and NVidia, but no complex benchmark is completely hardware neutral. This one apparently less than most. Personally I don't really care. Synthetic benchmarks are somewhat useful for reviewers, but for us gamers they're pretty pointless.

Some of you can run the basemark benchmark on Windows with Vulkan VS Linux Vulkan ? I really wish to see the score of Vulkan between Windows and Linux on that kind of benchmark. Please

Last edited by Vinouch on 2 September 2020 at 5:38 pm UTC

Quoting: The_AquabatPhoronix is reporting that Mesa 20.3 should come with a big boost on Basemark for Navi

Did he test any of the Mesa releases between 20.0.4 and 20.3? I expect the progress might be more gradual.

Quoting: The_Aquabat

Quoting: tuubi

Quoting: The_AquabatPhoronix is reporting that Mesa 20.3 should come with a big boost on Basemark for Navi

Did he test any of the Mesa releases between 20.0.4 and 20.3? I expect the progress might be more gradual.

It's ubuntu 20.04.1 (not Mesa) I think that comes with stock Mesa 20.0.8. But yes maybe Mesa 20.2 should have been tested.

Yeah sorry, of course I meant 20.0.8. Don't know why I typed a 4 even after I double checked the version in the repo.

Posting some results here, since I did in the other thread.

CPU: Ryzen 2700X
GPU: RTX 2080 Ti



OpenGL: 12,653



Vulkan: 13,164



Sorry it's broken for you, Aquabat.