34-Way Graphics Card Comparison On Ubuntu 17.10

[phoronix]

Phoronix: 34-Way Graphics Card Comparison On Ubuntu 17.10

As part of marking AMD's open-source driver strategy starting 10 years ago, among other articles, over the past week I posted an 18-way Radeon graphics card comparison on Ubuntu 17.10 while upgrading to the latest drivers. Taking those numbers further and putting them into more perspective, here is now a brief 34-way comparison with the NVIDIA GeForce counterparts added in.

http://www.phoronix.com/vr.php?view=25454

[pinguinpc]

Michael thanks for 1080p tests however fault more games

This article shows amd need improve in 1080p

[Michael]

Michael thanks for 1080p tests however fault more games

This article shows amd need improve in 1080p

The selection of games was very limited by what could still run on old GPUs and still show scaling with newer GPUs.

[M@GOid]

Michael

Are you having problems with F1 2017 and RAVD too? Here the game is crashing all the time, no matter if I try Mesa 17.2, 17.3 or 17.4 on a RX 570. Other Vulkan titles run fine, thought.

[Michael]

Michael

Are you having problems with F1 2017 and RAVD too? Here the game is crashing all the time, no matter if I try Mesa 17.2, 17.3 or 17.4 on a RX 570. Other Vulkan titles run fine, thought.

F1 2017 and RADV are working out fine for me, except for hangs when trying GCN 1.0 with AMDGPU and on Vega there is a few corruption issues. Using Linux 4.14 + Mesa 17.4.

[Ophidian]

Where are the numbers for the RX480? It is listed as one of the tested cards but does not appear in any graph

[Michael]

Where are the numbers for the RX480? It is listed as one of the tested cards but does not appear in any graph

It's been running into problems lately of losing the display output during testing.

[Michael]

No R9 290/390 GPUs?

I don't have any R9 390. My R9 290 system is busy in a 4U rackmounted case.

[Ophidian]

It's been running into problems lately of losing the display output during testing.

It would probably be a good idea to mention this in the article or not include it in the list of what was tested

[tildearrow]

Typo:

With Tesseract these numbers are slightly better for RadeoNSI but